How to Pronounce つ (tsu) in Japanese: The Voiceless Alveolar Affricate English Lacks

Pronouncing つ (tsu) in Japanese comes down to one sound English never uses at the start of a word: the voiceless alveolar affricate [t͡s], followed by the compressed Japanese vowel [ɯ].¹² The same cluster appears at the end of "cats" and "klutz". Your mouth already knows how to make it, but your brain has to allow it in a position English phonotactics forbids.³

Overview

What つ encodes

つ is one kana, one mora, and one consonant-plus-vowel syllable.¹⁴ The consonant is the affricate [t͡s], a single sound written with a tie bar. The vowel is the Japanese close back compressed [ɯ̟ᵝ], not the rounded English "oo".¹²

A mora is the basic timing unit of Japanese. In a mora-timed analysis, every mora is ideally equal in length.⁴⁵ One つ is therefore one beat, no more.

月つきがきれいですね。
"The moon is beautiful, isn't it."

The same kana shape appears at a smaller size as っ (the sokuon).⁶ The small form is a separate mora that marks a geminate consonant, a doubled consonant. It is a one-mora period of silence rather than a CV unit.⁶

つ and っ are different morae, not one kana in two sizes

The full-size つ is the CV mora [tsɯ]. The small っ is a silent beat that doubles the next consonant. They share a shape on the page; they do not share a job.⁶

Why this sound earns its own article

English permits the [t͡s] cluster only at the end of a syllable, in words like "cats", "hats", and "klutz".³ It is absent from the English onset inventory, the set of sounds allowed at the start of a syllable. That is why "tsunami" often comes out in English as "soo-nami".⁷³

The articulation is not a new mouth movement. English speakers already produce [t͡s] in coda position, at the end of a syllable. The work is to allow the cluster at the onset of a syllable, where English phonotactics blocks it.⁷³

The phoneme: a voiceless alveolar affricate

IPA: [t͡s] plus [ɯ]

The IPA symbol is ⟨t͡s⟩, or sometimes ⟨t͜s⟩. The tie bar signals a single segment, though it may be omitted, yielding ⟨ts⟩.⁸ In plain terms, an affricate is a stop that releases as a fricative, all in one beat.

The reference definition describes how the sound is made. First, the airflow stops completely. Then the tongue directs the air to the sharp edge of the teeth, causing high-frequency turbulence.⁸ The place of articulation is alveolar: the blade of the tongue sits at the alveolar ridge, the gum line just behind the upper teeth.⁸

The vowel that follows in つ is [ɯ̟ᵝ], compressed rather than rounded.² The lips do not pucker the way they would for English "oo".²

妻つまと一緒いっしょに住すんでいます。
"I live together with my wife."

The t-row allophone story

The five t-row kana た, ち, つ, て, と all spell a single phoneme /t/. The surface pronunciation shifts with the following vowel.² The pattern depends on position and is exceptionless in standard Tokyo speech.

Kana	Phoneme + vowel	Surface IPA	Hepburn	Kunrei
た	/ta/	[ta]	ta	ta
ち	/ti/	[tɕi]	chi	ti
つ	/tu/	[tsɯ]	tsu	tu
て	/te/	[te]	te	te
と	/to/	[to]	to	to

Before /a/, /e/, and /o/, the phoneme surfaces as plain [t].² Before /i/, it palatalizes to the alveolo-palatal affricate [tɕ], so ち is [tɕi]. The standard analysis treats alveolo-palatal [tɕ dʑ ɕ ʑ] as positional allophones of /t d s z/ before /i/.² Before /u/, /t/ surfaces as the alveolar affricate [ts], so つ is [tsɯ].²

This three-way distribution is why romanization systems disagree on the row. Kunrei-shiki, established by Japanese Cabinet Order No. 1 of 29 December 1954 and standardized as ISO 3602:1989, writes the row by kana position: ta-ti-tu-te-to.⁹ Modified Hepburn writes the row by pronunciation: ta-chi-tsu-te-to.¹⁰

A learner reading older textbooks, government documents, or signage may meet つ written as "tu".⁹¹⁰ The sound is still [tsɯ]; the spelling is a notation choice, not a different pronunciation.⁹¹⁰

Articulation: a step by step walkthrough

Step 1: tongue position for [t]

Place the tip of the tongue against the alveolar ridge, just behind the upper front teeth, and block the airflow completely.⁸ This is the same closure English uses for the [t] in "stop".⁸

The closure is brief; nothing escapes the mouth during this phase.⁸

Step 2: release into [s]

Relax the tongue tip a small amount so that the trapped air escapes as a hiss [s]. That single release is the fricative part of the affricate.⁸

The stop and the fricative are one segment, transcribed with a tie bar [t͡s], not two consecutive sounds.⁸ Tokyo University of Foreign Studies's pronunciation course frames the rule as "it shouldn't be [ts]+[u]" pronounced as two beats.¹¹

Drill from [s] to [t͡s] without changing mouth shape

Start by saying [su]. Then move the tongue to the back of the upper front teeth to pronounce [tsu], while keeping the lips and jaw in the same position.¹¹ Build through [su-tsu], then [sutsu], then [tsu]. Keep going until the closure feels like part of the same motion as the hiss.¹¹

Step 3: add the vowel [ɯ]

Hold the lips neutral, not rounded.² The Japanese /u/ is a close near-back vowel with the lips unrounded [ɯ̟] or compressed [ɯ̟ᵝ]. It is never the protruded English "oo".²

The vowel and the affricate share one mora; the whole つ is one beat in the mora-timed rhythm of Japanese.⁴⁵

The "cats" shortcut

The cluster [t͡s] already exists in English at the end of a syllable. The end of "cats", "hats", "nuts", and "eats" is exactly [ts]. The English coda inventory lists /ts/ among stop-plus-fricative sequences such as /pθ/, /ps/, /bz/, /tθ/, /ts/, /dθ/, /dz/, /ks/, /gz/, with "klutz" as a textbook example.³

The exercise is to take that already-fluent coda [ts] and start a new syllable with it. Then add [ɯ]. Saying "cats" and then immediately "oo" with no pause between the words approximates つ.

次つぎは何なんですか。
"What's next?"

Why English speakers drop the t in "tsunami"

English permits /ts/ only at the end of a syllable

The English phonology reference lists the onset inventory of /s/ plus voiceless stop clusters (/sp/, /st/, /sk/) and excludes /ts/ from any onset list. /ts/ appears only in the coda inventory.³ The constraint is positional, not phonetic. The mouth can make the cluster (cats, klutz), but English phonotactics forbids it at the start of a word.³

Loanwords that begin with [t͡s] in their source language sit on the wrong side of that constraint. Russian tsar, Tswana tsetse, and Japanese tsunami all carry the cluster on paper. In casual English speech, they often surface as if the [t] were absent.⁷

What the brain does with a banned cluster

The standard English-speaker repair is documented directly for tsunami. Most English speakers alter the word's initial /ts/ to an /s/ by dropping the "t", since English does not natively permit /ts/ at the beginning of words. The original Japanese pronunciation is /ts/.⁷

The repair pattern is general. When an L2 onset is allowed in the source language but not in the listener's L1, the listener substitutes the nearest allowed onset.⁷³ For English /ts/-onset loans, that nearest shape is plain [s].⁷³

Cognates that expose the same gap

Tsar, tsetse, tsunami, tsukemen, and tsundere all carry the [t͡s] cluster on paper. English mouths tend to smooth it the same way they smooth the [t] out of tsunami.⁷

The Japanese fix is to refuse the repair

Once the pattern is visible, the work for Japanese is to override the English-side reflex on purpose. Keep the [t], make the alveolar closure, release into [s], and add [ɯ].⁸¹¹ The repair is a fact about English phonotactics, not about Japanese.⁷³

Common substitutions and how to fix them

つ pronounced as す (su)

The error is dropping the [t] entirely, so that つ is realized as [sɯ], the same sound as the kana す.¹¹¹² The cause is English phonotactic repair carrying over into Japanese. The English mouth defaults to a continuous hiss because that is the shape English allows at the start of a word.⁷³

The fix is to start every つ from a real alveolar stop, not from a sibilant.¹¹ The drill from Tokyo University of Foreign Studies is explicit: the stop and release are one motion, but the stop must actually happen.¹¹

A minimal-pair check exposes the error directly:

机つくえ
"desk"

救すくえ
"rescue (imperative of 救う)"

If the listener hears no difference between these two, the [t] is being dropped.¹¹

つ pronounced as ちゅ (chu)

The error is realizing つ as [tɕɯ], the alveolo-palatal affricate that belongs to ち plus the contracted sound (yōon) ゅ, not the alveolar affricate that belongs to つ.² The cause is acoustic proximity. The two affricates are close neighbors. The t-row allophone pattern can pull untrained ears toward [tɕ], the more familiar shape to English speakers who already use it in "church".²

The fix is tongue position. Keep the tip forward on the alveolar ridge.⁸ The palatalized [tɕ] requires the tongue blade to rise toward the hard palate. The alveolar [ts] does not.⁸²

つ pronounced as "t-su" in two beats

The error is splitting [t͡s] into a stop plus a separate fricative. This produces what sounds like two morae where there should be one.¹¹ The cause is treating "tsu" as if it were spelled out, first a /t/ and then a /s/, instead of as a single affricate. The tie bar in the IPA notation [t͡s] is the whole point.⁸

The fix is rhythmic, not articulatory. One release, one mora.⁴ Japanese rhythm is mora-timed, and each mora is supposed to take an equal amount of time. A two-beat つ stretches the word out of shape and changes the rhythm the listener parses.⁴⁵

A two-beat つ rewrites the word's rhythm

つき as "tsu-ki" in two morae is correct. つき as "t-su-ki" in three morae has a different rhythm. To a native ear, it sounds like a longer, mis-shaped word, not just a foreign accent.⁴⁵

つ in context: minimal pairs and high-frequency words

つ vs す minimal pairs

The clearest way to test the [t] of つ is to put it next to the matching す word and listen for contrast only at the onset.¹¹ The vowel is identical in each pair. Only the affricate vs. the fricative onset differs.²

つ word	Meaning	す pair	Meaning
つくえ (机)	desk	すくえ (救え)	rescue! (imperative)
つき (月)	moon, month	すき (好き / 鋤)	liking, or plow
つな (綱)	rope	すな (砂)	sand

月つきを見みてください。
"Please look at the moon."

好すきな食たべ物ものは何なんですか。
"What food do you like?"

High-frequency つ words

A short list of common words makes つ unavoidable in the first week of many beginner courses. These words appear early in introductory textbooks, and each is a useful drill target.

• 月 tsuki (moon; also getsu or gatsu in the "month" readings).
• 机 tsukue (desk).
• 妻 tsuma (wife).
• 次 tsugi (next).
• 続く tsuzuku (to continue); the second mora is づ, the historically voiced counterpart of つ, discussed below.¹³

The Tokyo University of Foreign Studies course also pairs word-initial examples (tsuki, tsukue) with word-medial examples (futsū "ordinary", itsuka "someday") and word-final examples (mittsu "three things", ikutsu "how many"). This helps the learner hear the same affricate in every position.¹¹

今日きょうは普通ふつうの日ひです。
"Today is an ordinary day."

いくつありますか。
"How many are there?"

Good to know

Treating っ as a small つ that gets pronounced

The regular つ is a CV mora [tsɯ]. The small っ is a different mora that marks a geminate consonant, or at the end of an utterance, a glottal stop.⁶ The two are written with the same kana shape at two sizes, and the difference is easy to miss on the page.

A learner who reads 切手 as ki-tsu-te with three morae and an audible つ has misread the small form. The word has two audible kana morae plus one silent mora of consonant doubling, written:

切手きって⁶
"postage stamp"

づ and ず are identical in standard Tokyo speech

Standard Tokyo Japanese has collapsed the historical four-way distinction of the yotsugana (じ／ぢ／ず／づ) to two. The contrast between [dz dʑ] and [z ʑ] has been lost in the Tokyo-based standard, though not in all regional varieties.¹³ In writing, づ is kept only when a word shows sequential voicing (rendaku) as a result of compounding, as in 神無月 kannazuki from 月 tsuki.¹³¹⁴

This is why 続く tsuzuku is spelled with づ rather than ず: the second mora is historically the voiced counterpart of つ, preserved in the spelling.¹³

続つづきを読よみたい。
"I want to read the continuation."

The "tu" spelling for つ in older or government materials

Kunrei-shiki, established by Japanese Cabinet Order No. 1 of 29 December 1954 and standardized internationally as ISO 3602:1989, romanizes the t-row by kana position: ta-ti-tu-te-to.⁹ A 2025 cabinet notification reversed the policy and made Hepburn the new standard. Even so, Kunrei spellings still appear on older signage, in linguistics papers, and in some Japan-internal documents.⁹¹⁰

The pronunciation is [tsɯ] regardless of the spelling.²¹⁰ The "tu" of Kunrei and the "tsu" of Hepburn point at the same sound.

The "eight suits" elision mnemonic

Saying "eight suits" quickly collapses the word boundary into roughly [eɪt͡suːts]. The middle of that phrase is a word-initial [ts] that English never produces on its own.¹⁵ The mnemonic borrows the allowed coda [ts] of "eight" and reuses it as the onset of the next syllable. That is exactly the move the learner needs for つ.³¹⁵

Retire the mnemonic once the affricate becomes automatic. If you lean on it too long, the vowel can slide into English [u] rather than Japanese [ɯ]. Then fixing the vowel becomes harder than building the consonant was in the first place.²

Pronouncing "tsunami" with a silent t when speaking Japanese

The English-speaker default for word-initial /ts/ is to drop the [t], since English does not natively permit /ts/ at the beginning of words.⁷ When the same speaker switches to Japanese, that reflex can carry over. It can turn every word-initial つ into す, including in 津波 itself.

The fix is to recognize the reflex as a phonotactic repair (a fact about English, not about Japanese) and override it on purpose:

津波つなみ⁷
"tsunami (said as [t͡sɯnami], not [sɯnami])"

See also

• How to Pronounce ふ (fu) in Japanese: The Voiceless Bilabial Fricative English Lacks
• The Japanese ら-Row (ら り る れ ろ): The Alveolar Tap That Is Neither R Nor L
• Long vs. Short Vowels in Japanese: The Distinction Beginners Miss
• Why "Tokyo" Is Two Syllables in English and Four Morae in Japanese: Loanwords as a Timing Drill
• Common Romaji Mistakes That Mislead Pronunciation
• Japanese Vowel Devoicing: Why です Sounds Like "Des"

References

Footnotes

1. Vance, Timothy J. The Sounds of Japanese. Cambridge University Press, 2008. ISBN 978-0-521-61754-3. https://www.cambridge.org/core/books/sounds-of-japanese/ ↩ ↩² ↩³

2. "Japanese phonology." Encyclopedia entry, summarizing standard Tokyo Japanese segmental inventory. https://en.wikipedia.org/wiki/Japanese_phonology ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶

3. "English phonology." Encyclopedia entry, onset and coda cluster inventory for General American and Received Pronunciation English. https://en.wikipedia.org/wiki/English_phonology ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹²

4. "Mora (linguistics)." Encyclopedia entry, definition of the mora and the Japanese mora-timed analysis. https://en.wikipedia.org/wiki/Mora_(linguistics) ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶

5. Kubozono, Haruo (ed.). Handbook of Japanese Phonetics and Phonology. Handbooks of Japanese Language and Linguistics, vol. 2. De Gruyter Mouton, 2015. ISBN 978-1-61451-252-3. https://www.degruyterbrill.com/document/doi/10.1515/9781614511984/html ↩ ↩² ↩³ ↩⁴

6. "Sokuon." Encyclopedia entry on the small っ (geminate-consonant mora). https://en.wikipedia.org/wiki/Sokuon ↩ ↩² ↩³ ↩⁴ ↩⁵

7. "Tsunami." Encyclopedia entry, English pronunciation note on word-initial /ts/ simplification. https://en.wikipedia.org/wiki/Tsunami ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹

8. International Phonetic Association. "Voiceless alveolar affricate." Reference article. https://en.wikipedia.org/wiki/Voiceless_alveolar_affricate ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹²

9. 文化庁 (Agency for Cultural Affairs). 訓令式ローマ字 (Kunrei-shiki rōmaji), Cabinet Order No. 1, 29 December 1954; ISO 3602:1989 Documentation – Romanization of Japanese. Summary reference: https://en.wikipedia.org/wiki/Kunrei-shiki_romanization (Cabinet status corrected by Cabinet notification of 22 December 2025 making Hepburn the new standard). ↩ ↩² ↩³ ↩⁴ ↩⁵

10. Hepburn, James Curtis. A Japanese and English Dictionary; with an English and Japanese Index. Shanghai: American Presbyterian Mission Press, 1867. (Origin of the Hepburn romanization system; "tsu" spelling for つ.) Modern reference summary: https://en.wikipedia.org/wiki/Hepburn_romanization ↩ ↩² ↩³ ↩⁴ ↩⁵

11. 川口義一・木下直子 (Yoshikazu Kawaguchi and Naoko Kinoshita, eds.). 「日本語の発音」(Online Japanese Pronunciation Course), Tokyo University of Foreign Studies / 早稲田大学日本語教育研究センター. Consonant module 2: "Let's master the pronunciation of Japanese [tsu]." https://www.japanese-pronunciation.com/eng/movie/consonant2 ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹

12. Wasabi Japanese Lessons. "Japanese Consonants: How to Pronounce つ." https://www.wasabi-jpn.com/magazine/japanese-lessons/japanese-consonants-how-to-pronounce-ts (limitation: language-school blog; used only for articulation walkthrough phrasing.) ↩

13. "Yotsugana." Encyclopedia entry on the four-kana じ／ぢ／ず／づ merger in standard Tokyo Japanese. https://en.wikipedia.org/wiki/Yotsugana ↩ ↩² ↩³ ↩⁴

14. "Rendaku." Encyclopedia entry on sequential voicing in compound morphemes. https://en.wikipedia.org/wiki/Rendaku ↩

15. Dr. Moku. "How To Pronounce the Japanese Syllable tsu." https://drmoku.com/blog/how-to-pronounce-the-japanese-syllable-tsu (limitation: language-school blog; used only for the "eight suits" mnemonic, which is its native contribution to the public domain.) ↩ ↩²