The Small つ (Sokuon): How to Read and Pronounce the Geminate Consonant

The small tsu (sokuon) in Japanese is a half-size kana. It is written っ in hiragana and ッ in katakana. It occupies one full beat of silence and doubles the consonant of the kana that follows it.¹² On the page, it has no sound of its own. In speech, you hear it as a held closure or a sharp cutoff. It is the difference between 来て kite "come" and 切手 kitte "stamp".³⁴

Overview

Sokuon is one of three "special morae" (特殊拍 tokushuhaku) in Japanese. The other two are the moraic nasal ん and the length mora that produces long vowels.¹⁵⁶ Each counts as one full beat in the rhythm of a word, even though none is a normal consonant-plus-vowel syllable.

The small つ is silent in itself. The mora it represents is filled by the closure of the consonant that comes after it, or, at the end of an utterance, by a glottal stop.¹⁷

What "sokuon" means

The technical name is 促音 (sokuon), from 促 "press, urge" and 音 "sound".²³ The same kana is called 小さいつ (chiisai tsu) or 小さなつ (chiisana tsu), "small tsu", in everyday speech and in most beginner textbooks.³

In linguistic notation, the sound is the moraic obstruent, written /Q/.¹⁵ Here, "obstruent" means a consonant made by blocking or narrowing the airflow. /Q/ has no place or manner of its own. It assimilates fully to the following obstruent, which produces the audible "doubling".¹⁷

Sokuon is pre-N5 foundational mechanics

The JLPT (Japanese-Language Proficiency Test) publishes no can-do statement for sokuon because it sits below the grammar levels, in the kana writing system itself.³⁵ Even so, several N5 words depend on it directly: 学校 gakkō, 切手 kitte, 待つ matsu, and 来る kuru all carry or generate sokuon in normal use.⁸

Where the small つ sits among the hiragana extensions

After the 46 base hiragana, a learner meets three classes of "modifier" kana. The small つ is the third of them.

Modifier class	Mark or kana	Effect on the base kana	Example
Dakuten / handakuten	゛ / ゜	Voices the base consonant	か → が, は → ぱ
Yōon (拗音)	small ゃ・ゅ・ょ	Fuses a glide onto an i-row kana	き + ゃ → きゃ
Sokuon (促音)	small っ	Holds one silent mora and doubles the next consonant	か + っ + た → かった

The Hiragana Chart article covers the 46 base kana the reader is assumed to know. Dakuten and Handakuten covers the voicing marks. This article covers the third class. Unlike full-size つ, which is the syllable tsu /tsɯ/, small っ has no sound of its own and signals gemination of whatever follows.¹³

The katakana counterpart ッ behaves identically and follows the same rules. It is the form used in loanwords (ベッド, ポッキー) and onomatopoeia.³

A note on names

文化庁 (the Agency for Cultural Affairs), NHK, and Japanese dictionaries use 促音 in technical writing.²⁹⁴ Beginner textbooks and IME (input method editor) help pages use 小さいつ or 小さなつ.³ Both refer to the same kana, and a learner will encounter both terms.

How the small つ is written

Size, position, and the katakana counterpart ッ

The 1986 Cabinet Notification 『現代仮名遣い』 directs that the つ used for sokuon should be written small: 「促音に用いる「つ」は，なるべく小書きにする。」² In practice, "small" means roughly half the height of a full-size kana. The glyph is otherwise identical to つ or ツ.³

In horizontal text the small kana sits on the baseline at the bottom-left of its cell. In vertical text it sits at the top-right of its cell.³ The katakana ッ obeys the same orthographic rule.²³ Stroke order for the half-size form matches the full-size つ; see Hiragana Stroke Order for the underlying strokes.

Typing っ on a Japanese IME

The standard way to enter っ is to type the doubled consonant of the following kana. The IME sees the doubled letter and inserts a small っ before the next syllable.

What you type	What you get
kitte	きって
gakkou	がっこう
ippon	いっぽん
matte	まって

For a standalone っ, use the prefix shortcut xtu or ltu. This is useful when there is no following kana, or when you need to override the IME. Some IMEs also accept xtsu and ltsu.³

Mistyping a single consonant drops the sokuon entirely

Typing kite instead of kitte yields きて, "come", not 切手 "stamp"; the IME has no way to guess that a double was intended.³ When a typed word reads wrong, check the consonant count before suspecting anything else.

How the small つ is pronounced

The silent-beat model: っ is a mora of held breath

Japanese is a moraic language: every kana is one mora, and morae are roughly equal in duration. The small つ is also exactly one mora, even though no sound is produced during it.¹⁵⁶

Phonetically, the speaker prepares the next consonant early and holds the closure for the duration of one mora before releasing into the following vowel.¹⁷ For a stop like /t/, that means the tongue is pressed against the alveolar ridge and held silent. For /s/, the constriction is held, and a hiss may be heard for the whole beat.

This silent beat is what English-trained ears tend to miss. There is no audible consonant to produce. There is only a measurable beat of held closure or sustained noise.¹ A useful cross-linguistic analogy is the middle of English clusters like "hot tea" or "bookkeeper": the closure between the two halves is roughly comparable to the Japanese geminate.¹

The diagram below shows the silent beat sitting as its own mora between two ordinary morae:

The first path is まって matte "wait": three morae, with the middle one silent. The second path is まつ matsu "pine": two morae, with no silence in either.¹⁴

What "geminate" means and why romaji doubles the consonant

A geminate consonant is one held for about twice the duration of its singleton counterpart. Acoustic studies of Japanese report closures on the order of two to three times the singleton length, depending on the consonant and the corpus.¹⁷

Hepburn romanization marks gemination by doubling the following consonant letter: kk, pp, tt, ss. The convention dates to James Curtis Hepburn's 1867 Japanese and English Dictionary. Modified Hepburn, the form used on Japanese road signs, passports, and in this article, preserves it.¹⁰¹¹

The canonical contrast pair is the one a learner will see in the first textbook chapter that introduces sokuon:

来きて³
"(please) come"

切手きって³⁴
"postage stamp"

The small っ is the only difference. The mouth shape for the /k/ and /t/ is identical in both words. The second word adds one mora of held closure between them.

The four common romanization patterns: kk, pp, tt, ss

In native and Sino-Japanese vocabulary, sokuon attaches to one of four consonant rows. Each row gets a doubled letter in Modified Hepburn.

っ + row	Romaji pattern	Example word	Romaji
か行 (k-row)	kk	学校	gakkō
ぱ行 (p-row)	pp	一本	ippon
た行 (t-row)	tt	待って	matte
さ行 (s-row)	ss	喫茶店	kissaten

All four readings are sourced from standard Japanese dictionaries.¹⁰⁴

学校がっこうに行いく。⁴⁸
"I go to school."

切符きっぷを買かう。⁴⁸
"I buy a ticket."

夫おっとは家いえにいます。⁴⁸
"My husband is at home."

喫茶店きっさてんで会あいましょう。⁴⁸
"Let's meet at the coffee shop."

For the palatalised yōon series, only the leading consonant doubles. For example, っ + し becomes ssh as in 一生 isshō "a whole lifetime". っ + つ is itself ungrammatical in native vocabulary.¹⁰¹¹

The tch exception for っち

Modified Hepburn writes っ + ち as tch, not cch: 抹茶 matcha, 一致 itchi, こっち kotchi.¹⁰¹¹

The rule reflects two facts. First, ち is romanised chi (not ti). Writing cchi would obscure the affricate sound. Second, the digraph tch was already established in English spelling (matchstick, kitchen) as a marker of geminated /tʃ/. That makes the choice consistent with the reader's existing spelling intuition.¹⁰¹¹

抹茶まっちゃを飲のむ。⁴⁸
"I drink matcha."

一致いっちしました。⁴⁸
"They matched."

Nihon-shiki and Kunrei-shiki write tti, not tch

The Japanese government's own romanization standards (Nihon-shiki and Kunrei-shiki) use ti for ち. They therefore render 抹茶 as matti and 一致 as itti.¹¹ On consumer products, place names, and most English-language signage in Japan, the spelling the reader actually sees is Modified Hepburn's matcha and Hatchōbori 八丁堀.

Word-final っ: the hard stop

When っ appears at the end of an utterance, there is no following consonant for it to assimilate to. The moraic obstruent /Q/ is then realised as a glottal stop [ʔ]: an abrupt closure of the vocal folds that cuts the preceding vowel short.¹³⁷

あっ³
"Oh!" (sudden realisation)

やめろっ³
"Stop it!" (emphatic command)

This use is conversational and stylistic. It is common in manga, dialogue, and informal text. It does not appear in formal prose, business writing, or official documents.³

A learner reading manga or social-media text will see a third possibility too: a small っ used between syllables for emphasis (すっごい suggoi for emphatic sugoi).³ The decision tree below sorts the three cases:

Where the small つ can and cannot appear

Native Japanese rules: only before voiceless obstruents

In native Japanese words (和語 wago) and Sino-Japanese words (漢語 kango), sokuon is restricted to the voiceless obstruents /p t k s/ and their palatal allophones /tɕ ɕ/ (ち and し).¹³⁷

Several restrictions follow from this:

• っ cannot start a word.¹³
• っ cannot precede a vowel.¹³
• っ does not precede /n m r w y/.³⁷
• っ does not precede the voiced obstruents /g z d b/; gemination of voiced consonants is essentially absent from the native lexicon.¹³⁷

Historically, many of these geminates arose from onbin (音便), sound changes in Late Old and Early Middle Japanese. In these changes, consonant-final morae assimilated to a following voiceless obstruent. The verbal te-form 待って comes from earlier 待ちて through exactly this process.¹²

Loanword and emphatic exceptions

Loanwords (外来語 gairaigo) openly break the voiceless-only restriction. The Japanese sound system imports the source word's geminate structure, even when the resulting cluster would be impossible in native vocabulary.

Pattern	Example	Romaji	Gloss
っ + d	ベッド	beddo	"bed"
っ + g	バッグ	baggu	"bag"
っ + z	グッズ	guzzu	"goods"
っ + h (fricative)	バッハ	Bahha	"Bach"

Readings sourced from standard reference.³

These voiced and fricative geminates are absent or vanishingly rare in native vocabulary. The loanword pattern is the only systematic source.¹³

Emphatic and dialectal speech can also allow unusual っ + voiced sequences in casual writing (e.g. すっごい for emphatic sugoi). These spellings reflect expressive lengthening rather than lexical gemination.³

Quick reference: the licensed and unlicensed following kana

Following kana	Native vocabulary	Loanwords
か行, さ行, た行, ぱ行 (and yōon きゃ・しゃ・ちゃ・ぴゃ etc.)	licensed	licensed
が行, ざ行, だ行, ば行	unlicensed	licensed
あ行 (vowels)	unlicensed	unlicensed
な行, ま行, ら行, わ行, や行	unlicensed	rare / transliteration only
は行	unlicensed (rare onbin exceptions)	licensed in ハハ-type transliterations (バッハ)

Coverage is based on the phonotactic rules, meaning sound-combination rules, summarised in the linguistic literature.¹³

Minimal pairs and ear training

The four pairs every beginner should drill

The doubled consonant changes the meaning. The silent beat is the only audible difference. Each pair below is everyday vocabulary, with both members appearing in standard reference dictionaries.⁴⁸

Without っ	Romaji	Gloss	With っ	Romaji	Gloss
来て	kite	"come" (te-form)	切手	kitte	"stamp"
過去	kako	"past"	括弧	kakko	"parentheses"
音	oto	"sound"	夫	otto	"husband"
部下	buka	"subordinate"	物価	bukka	"prices"

For the drill, say each pair aloud and count morae as you go: ki – te (two beats) versus ki – [hold] – te (three beats). The contrast is timing, not loudness.

Each pair appears in a natural sentence below, so the silent beat sits inside ordinary prose rather than in isolation.

来きて、切手きってを買かってきて。⁴⁸
"Come here, then go buy stamps and come back."

過去かこのことより、括弧かっこの中なかを見みてください。⁴⁸
"Never mind the past; please look inside the parentheses."

あの音おとは、夫おっとの足音あしおとです。⁴⁸
"That sound is my husband's footsteps."

部下ぶかに物価ぶっかのことを聞きかれた。⁴⁸
"My subordinate asked me about prices."

A fifth pair, saka 坂 "slope" versus sakka 作家 "writer", works the same way:

坂さかの上うえに作家さっかが住すんでいる。⁴⁸
"A novelist lives at the top of the slope."

Small っ versus full-size つ

Full-size つ is a syllable pronounced [tsɯ]. Small っ has no sound of its own.¹³

まつ "pine tree" is two morae: ma + tsu. まって "wait (te-form)" is three morae: ma + [held closure] + te.¹⁴ The first word has a normal [ts] onset on its second mora. The second word has a silent mora in the middle and a [t] release into the third. See the timing diagram in the silent-beat-model section above for the contrast in picture form.

The same contrast scales up to longer words. がっこう is four morae (ga – [held closure] – ko – [length]), while the hypothetical がこう would be three (ga – ko – [length]).¹⁵ Both the held closure and the vowel length are full morae. The writing system marks each with its own kana.

Length contrasts come in two flavours, and they are not the same thing

Beginner ears often misfile any length contrast as "the small tsu thing". The pair おじさん ojisan "uncle" versus おじいさん ojiisan "grandfather" is a long-vowel contrast (an added い mora), not a sokuon contrast.¹⁵⁴ Sokuon lengthens consonants only. Vowel length is marked with a vowel kana, or with ー in katakana.

Good to know

Undershooting the silent beat in きて versus きって

English speakers tend to merge the silent mora into the following consonant, producing a single-length kite when 切手 kitte "stamp" was meant. The fix is to count morae out loud (ki – [hold] – te), not to make the /t/ louder. The difference is timing, not volume.¹⁷ The corrected form is:

切手きって⁴⁸
"postage stamp"

Confusing sokuon with a long vowel in おじいさん

Writing おっじさん for "grandfather" applies the small tsu where a long vowel is needed. Sokuon lengthens consonants only. Vowel length is marked by a vowel kana (or ー in katakana). The correct form is:

おじいさん⁴
"grandfather"

The two contrasts are independent and both contribute to Japanese being a moraic language.¹⁵

Writing っ before a vowel or a voiced consonant in native words

Forms like 「あっい」or 「おっが」 are ungrammatical in native vocabulary. Native sokuon only appears before /p t k s/ and their palatal variants.¹³ The fix is to drop the っ. Loanwords (ベッド, バッグ) are the licensed exceptions to this rule, not a general permission for native words.¹³

Word-final っ in formal writing

The emphatic glottal-stop っ (あっ, やめろっ) belongs to casual dialogue, social-media text, and manga. Formal prose, business correspondence, and academic Japanese omit it.³ Reach for it only when the register of the surrounding text already accepts it.

Small tsu is a held breath

A one-line mnemonic worth repeating aloud: small tsu means hold the next sound for one beat, then release. This framing maps cleanly onto a body sensation a learner can rehearse: set the mouth for the next consonant, hold for one beat, release. It also separates っ from its neighbours in the modifier system: dakuten adds a voicing, yōon fuses a glide, and the long vowel extends a vowel.⁹

Why romaji uses tch for っち

Modern Hepburn descends from James Curtis Hepburn's 1867 dictionary, which adopted English spelling conventions. English already used tch to spell geminated /tʃ/ (kitchen, matchstick), so 抹茶 became matcha rather than maccha.¹⁰¹¹ The Japanese government's own romanization standards (Nihon-shiki, Kunrei-shiki) use ti for ち and therefore matti for 抹茶. The spelling readers actually encounter on products and signs is Modified Hepburn's matcha.¹¹

Small kana are a 20th-century standardisation

Before the post-war orthography reforms, sokuon was often written with a full-size つ, and the reader inferred gemination from context. The explicit half-size kana became the standard with the 1946 cabinet directive 「現代かなづかい」. The 1986 『現代仮名遣い』 (内閣告示第1号) reaffirmed it and states that つ used for sokuon should be written small.² The same reform standardised the small や・ゆ・よ for yōon. Learners therefore encounter both conventions as a single 20th-century systematisation rather than as two separate quirks.²

Many native geminates come from onbin sound change

Forms like 待って (mat-te) and 買って (kat-te) arose historically from 待ちて (machi-te) and 買ひて (kahi-te) through 促音便 (sokuonbin). This was a Late Old and Early Middle Japanese assimilation that produced moraic obstruents at morpheme boundaries.¹² This is why the te-form of consonant-stem verbs is the single largest source of sokuon in everyday Japanese: every つ-row, う-row, and る-row verb's te-form contains one.¹²

Names and IME shortcuts at a glance

The kana goes by 促音 in technical writing and 小さいつ or 小さなつ in everyday speech.²⁹³⁴ To type it, double the consonant of the following kana (kitte → きって, gakkou → がっこう); to type a standalone っ, use xtu or ltu (some IMEs also accept xtsu and ltsu).³

See also

• The Japanese Consonant Inventory: Phonemes, Allophones, and the Kana Chart
• Common Romaji Mistakes That Mislead Pronunciation
• Japanese Pronunciation Drills: A Daily 5-Minute Protocol with Minimal Pairs, Shadowing, and Record-and-Compare
• Why "Tokyo" Is Two Syllables in English and Four Morae in Japanese: Loanwords as a Timing Drill
• Wago, Kango, Gairaigo, Konshugo: The Four Vocabulary Strata of Japanese

References

Footnotes

1. Vance, Timothy J. The Sounds of Japanese. Cambridge University Press, 2008. Chapter on moraic phonology and obstruent geminates (the /Q/ analysis). ISBN 9780521617543. ↩ ↩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

2. 文化庁. 『現代仮名遣い』(昭和61年内閣告示第1号). Government of Japan, Cabinet Notification No. 1, 1986. https://www.bunka.go.jp/kokugo_nihongo/sisaku/joho/joho/kijun/naikaku/gendaikana/honbun_dai1.html - Section "第1・5 促音", note: 「促音に用いる「つ」は，なるべく小書きにする。」 ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸

3. Wikipedia contributors. "Sokuon". Wikipedia, The Free Encyclopedia. https://en.wikipedia.org/wiki/Sokuon (limitation: secondary source; used only for points cross-confirmed by ¹ or ²). ↩ ↩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

4. 新村出 編. 『広辞苑』第七版. 岩波書店, 2018. Entries 切手・夫・音・一致・抹茶・学校・一本・待つ・作家・坂・部下・物価・括弧・過去 used to verify readings and meanings of the example vocabulary. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶ ↩¹⁷ ↩¹⁸ ↩¹⁹ ↩²⁰ ↩²¹ ↩²²

5. 国立国語研究所 (NINJAL). Entry on モーラ (mora) / 特殊拍 in the institute's reference materials and corpus documentation. https://www.ninjal.ac.jp/ ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷

6. Wikipedia contributors. "Mora (linguistics)". Wikipedia, The Free Encyclopedia. https://en.wikipedia.org/wiki/Mora_(linguistics) (limitation: secondary source; cross-confirmed by ¹ and ⁵). ↩ ↩²

7. Wikipedia contributors. "Japanese phonology". Wikipedia, The Free Encyclopedia. https://en.wikipedia.org/wiki/Japanese_phonology (limitation: secondary source; used only for points cross-confirmed by ¹). ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹

8. Jisho.org (electronic Japanese-English dictionary aggregating JMdict / EDICT / KANJIDIC; Electronic Dictionary Research and Development Group, Monash University). https://jisho.org/ (used as a cross-check for kana readings and JLPT tagging on the example words). ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴

9. NHK放送文化研究所, ed. 『NHK日本語発音アクセント新辞典』. NHK出版, 2016. Appendix: 発音解説 (sokuon, hatsuon, chōon as 特殊拍). ISBN 9784140113455. ↩ ↩² ↩³

10. Hepburn, James Curtis. A Japanese and English Dictionary; with an English and Japanese Index. Shanghai: American Presbyterian Mission Press, 1867. First print attestation of the romanization system later named after the author; basis of Modified Hepburn used today, including the doubled-consonant convention for sokuon and the tch digraph for っち. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶

11. Wikipedia contributors. "Hepburn romanization". Wikipedia, The Free Encyclopedia. https://en.wikipedia.org/wiki/Hepburn_romanization (limitation: secondary source for the doubled-consonant and tch conventions; primary attestation is ¹⁰). ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷

12. Frellesvig, Bjarke. A History of the Japanese Language. Cambridge University Press, 2010. Sections on the development of the moraic obstruent and onbin-derived geminates. ISBN 9780521653206. ↩ ↩² ↩³