How to Pronounce ふ (fu) in Japanese: The Voiceless Bilabial Fricative English Lacks
The Japanese fu sound, written ふ in hiragana, is a voiceless bilabial fricative [ɸ] followed by the close back compressed vowel [ɯ̟ᵝ]. Together, they are transcribed [ɸɯ̟ᵝ].12 English has no single consonant that matches it. Native ears hear neither English F nor English H, because the friction is made between the two lips alone.31
Overview
What ふ is, in one sentence
ふ is one kana representing one mora. It is realized as the voiceless bilabial fricative [ɸ] before the Japanese close back compressed vowel [ɯ̟ᵝ].12 In core vocabulary, [ɸ] occurs only before /u/. It can be analyzed as an allophone, or context-based variant, of the /h/ phoneme in that position.1
The vowel itself is unrounded or compressed rather than protruded: "a close near-back vowel with the lips unrounded [ɯ̟] or compressed [ɯ̟ᵝ]. When compressed, it is pronounced with the side portions of the lips in contact but with no salient protrusion."1
服を買いました。4
"I bought clothes."
Why English speakers hear "fu" and "hu"
English /f/ is labiodental: the upper teeth touch the lower lip, and air is forced through the narrow gap between them.3 Japanese [ɸ] differs from it in exactly one feature, place of articulation. That is why the English ear classifies [ɸ] as a soft F.32
English /h/ is glottal, made deep in the throat with no lip activity at all.1 ふ has clear lip activity, so a fully glottal [h] also sounds wrong to a native ear. Learners who hear "hu" in romanizations often miss that the lips are doing real work.1
The romanization itself is a notation choice, not a phonetic choice. Hepburn writes the kana as "fu" to capture the lip friction. Nihon-shiki and Kunrei-shiki write the same kana as "hu" to keep the は-row spellings systematic across the row.256
Where ふ sits on the kana chart
ふ belongs to the は-row, which spells a single phoneme /h/ in modern Japanese phonological analysis.1 The five kana in the row surface as three different consonants, depending on the following vowel: は [ha], ひ [çi], ふ [ɸɯ], へ [he], ほ [ho].1
ふ is the only modern descendant of the row's older [ɸ] articulation. The others lenited, or weakened, further to the glottal [h] before /a e o/ and the palatal [ç] before /i/.72 That history is why ふ sounds different from its row-mates and why it feels like an outlier at first.
風が強いです。8
"The wind is strong."
The articulation: lips only, no teeth
English [f]: lip touches teeth
English /f/ is a voiceless labiodental fricative: the upper front teeth meet the lower lip, and air is forced through the narrow gap between them.3 This is the default articulation for many English-speaking learners. The brain reaches for it whenever a romanization shows "f".3
The contrast with Japanese [ɸ] is a single feature: labiodental (teeth involved) versus bilabial (lips only).39 The manner (fricative) and the voicing (voiceless) are identical; only the place changes.
Japanese [ɸ]: lips alone, narrow gap
The Japanese fricative in ふ is produced by sending air through a narrow channel between both lips, with no teeth contact. The narrowing creates turbulence as the air passes through.3 A widely cited learner walkthrough phrases the gesture directly: "you are going to bring your lips close together and make a sound by causing friction as air passes through the chink between your lips."9
The IPA features of [ɸ] are voiceless (no vocal-fold vibration), bilabial (both lips), fricative (continuous airflow through a narrow channel), and pulmonic egressive (air from the lungs).3 The tongue stays neutral. The work is all in the lips.
The candle-blow drill
The standard teaching shortcut separates the bilabial lip setting from the vowel: "bring your lips together like you're blowing out a candle (no need to stick them out like you're kissing), then try to say 'fu'."10 This drill is widely cited in English-language teaching materials. It works because it moves the lips into the [ɸ] shape before the vowel arrives.910
Mime blowing out a candle gently, freeze the lip shape, and only then voice [ɯ] through the lip channel without changing the lip position. If the upper teeth join in at any point, the articulation has slipped back to English [f].10
二つください。11
"Two, please."
富士山に登りたい。12
"I want to climb Mt. Fuji."
Side-by-side: [f], [h], [ɸ]
This table shows what changes between the three fricatives English speakers reach for, and what stays the same. The manner and voicing are identical. Only the place of articulation shifts.
| Sound | Place | Manner | Voicing | What the lips do | What the tongue does |
|---|---|---|---|---|---|
| English [f]3 | Labiodental | Fricative | Voiceless | Upper teeth touch lower lip | Neutral |
| English [h]1 | Glottal | Fricative | Voiceless | No activity | Neutral |
| Japanese [ɸ]312 | Bilabial | Fricative | Voiceless | Both lips together, narrow gap | Neutral |
The bilabial [ɸ] is the standard pronunciation for ふ across registers in Tokyo Japanese. Intervocalic /h/, or /h/ between vowels, may be breathy-voiced [ɦ] rather than fully voiceless [h] when not preceded by a pause. This detail affects flowing speech around words like 富士, but not the segmental identity of the consonant itself.1
Why the は-row behaves this way
/h/ has three allophones
The /h/ phoneme surfaces as three different consonants depending on the following vowel: [h] before /a e o/, [ç] (a voiceless palatal fricative) before /i/, and [ɸ] before /u/.113 One phoneme has three surface sounds. The place of articulation tracks the vowel.
Japanese speakers treat all three as "the same consonant" because they never contrast in native vocabulary. The learner does not pick one freely; the following vowel decides which one appears.1
人がいます。1
"There is a person."
母は元気です。1
"My mother is well."
Historical note: /p/ to /ɸ/ to /h/
The reconstructed Old Japanese consonant inventory includes /p/ as the original initial of every modern は-row kana. "There is general agreement that word-initial p had become a voiceless bilabial fricative [ɸ] by Early Modern Japanese."7 The row then split: [h] before /a e o/, [ç] before /i/, and [ɸ] left over only before /u/.17
ふ is therefore a fossil. It kept the older [ɸ] articulation while its siblings moved on. That is why the kana shape, the spelling, and the row membership all reflect the consonant's history rather than its current sound.27 The trajectory is reconstructed from comparative evidence, written records, and 16th-century Portuguese missionary transcriptions. It is not contested in the standard literature.7
What this means for spelling and romanization
Hepburn romanization writes ふ as "fu" to track the surface bilabial fricative: ふ "is romanized fu in Hepburn romanization instead of hu as in Nihon-shiki and Kunrei-shiki rōmaji."2 Kunrei-shiki writes the row systematically as ha-hi-hu-he-ho, by kana position rather than surface sound. That matches the underlying /h/ phoneme.6
Kunrei-shiki was established by Japanese Cabinet Order No. 1 of 29 December 1954 and adopted internationally as ISO 3602:1989.6 A learner reading older textbooks, Japanese government documents, or older signage may meet ふ written as "hu". The sound is still [ɸɯ̟ᵝ], and the spelling is a notation choice, not a different pronunciation.256
On 16 December 2025, the Japanese government decided to issue a cabinet notification on 22 December revising national rules on romanization for the first time in about 70 years. The revision made Hepburn the national standard instead of Kunrei-shiki.56 The change is administrative. The underlying sound is unchanged.56
Common learner substitutions and how to fix them
Substitution 1: English [f] (lip on teeth)
The error is producing ふ with the labiodental contact of English /f/, so the upper teeth bite the lower lip on the way into the vowel.310 The cause is L1 default, meaning a habit carried over from your first language. English speakers already have a "voiceless fricative made with the lips" in their inventory, and the brain reaches for it. The teeth come along for the ride.3
The fix is the candle-blow drill: bring the lips together, keep the teeth out of the gesture, and then voice [ɯ] into the lip channel without changing the lip position.10 A useful self-check is 富士 Fuji. If it sounds like English "foo-jee" with audible teeth friction, the contact is labiodental and needs to smooth out to lips only.12
Substitution 2: English [h] (glottal puff)
The error is the opposite extreme: dropping all lip activity and producing a fully glottal [h], so ふ comes out as English "hoo".1 The cause is over-correction. A learner who has been told "ふ is not English F" sometimes lands on the other extreme and removes the lip friction entirely.110
The fix is to keep the lips active. The Wasabi learner walkthrough is explicit that "friction as air passes through the chink between your lips" is the defining gesture, not its absence.9
Substitution 3: Spanish or German [f]
The error is producing a labiodental [f] characteristic of Spanish, German, or French. In place of articulation, it is indistinguishable from English /f/.3 The same labiodental setting is the default for most European-language /f/ phonemes. As a result, L1 transfer from Spanish or German lands in the same wrong place as English-speaker transfer.3
The fix is the same candle-blow drill. The remedy is L1-independent because the target articulation is L1-independent.10
The fix, in three drills
Three small drills, run in sequence, cover both diagnosis and correction. None require recordings or partners. Each takes seconds.
| Drill | What to do | What it checks |
|---|---|---|
| Candle10 | Blow out an imaginary candle gently; freeze the lip shape; voice [ɯ] without moving the lips. | Forces bilabial gesture before the vowel arrives. |
| Mirror39 | Watch the lips while saying ふ; check whether the upper teeth touch the lower lip at any point. | Catches labiodental slips visually. |
| Row sweep113 | Say は・ひ・ふ・へ・ほ across the row out loud, holding the same /h/ identity. | Confirms lip activity appears at ふ, palatal at ひ, glottal at は・へ・ほ. |
福岡に住んでいます。14
"I live in Fukuoka."
布団で寝ます。15
"I sleep on a futon."
ふ in real words
Native words: 富士 (Fuji), 服 (fuku), 布団 (futon), 福岡 (Fukuoka)
ふ appears in week-one vocabulary, so learners get immediate practice in word-initial position across familiar nouns and a high-frequency place name. Each word below carries the same lip cue: keep the lips together, use no teeth, then release into the following vowel.
| Word | Reading | Meaning | Tag |
|---|---|---|---|
| 服4 | fuku | clothes | JLPT N5 |
| 二つ11 | futatsu | two things | JLPT N5 |
| 冬16 | fuyu | winter | JLPT N5 |
| 風8 | fū (in compounds) | wind | JLPT N5 |
| 布団15 | futon | Japanese bedding | JLPT N4 |
| 普通17 | futsū | ordinary | JLPT N4 |
| 富士山12 | Fujisan | Mt. Fuji | Common-word proper noun |
| 福岡14 | Fukuoka | city / prefecture | Common-word proper noun |
冬は寒いです。16
"Winter is cold."
普通の日です。17
"It's an ordinary day."
Loanwords back into English: tsunami, futon, Fuji
布団 futon entered English with the original [ɸ] flattened to labiodental [f] or omitted altogether in some speakers' productions.15 富士 Fuji entered English place-name vocabulary the same way. English mouths produce labiodental [f], not bilabial [ɸ], at the onset.12
The same import pattern applies to つなみ 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, though the original Japanese pronunciation is /ts/."18 The English-mouth repair is similar: foreign onsets get nudged to the nearest permitted English shape.
Why "Fuji" sounds wrong in English mouths
The mismatch is one feature, place of articulation: the English speaker substitutes labiodental [f] for bilabial [ɸ], producing audible teeth friction where the original has none.32 Native Japanese ears notice the teeth because the friction profile of [f] is acoustically distinct from [ɸ]. The teeth shape the air stream in a way the lips alone do not.39
The fix is not a new gesture. It is the removal of the teeth from the existing one. The lips already know how to make friction. They just need to do it without help.
ふ in katakana and the ファ・フィ・フェ・フォ system
フ as the base loanword sound
Katakana フ uses the same [ɸ] articulation as hiragana ふ. The consonant does not change when the script does.2 "The katakana フ is frequently combined with other vowels to represent sounds in foreign words," and those combinations form the extended-katakana system that Japanese uses to write English F-sounds for the first time.2
Small-vowel combinations for English F-sounds
A small vowel kana after フ writes the consonant [ɸ] before a vowel other than /u/. Each combination is one mora. The consonant in every one of them is the same bilabial [ɸ] used in native ふ. Only the following vowel changes.12
| Kana | Reading | Example | Meaning |
|---|---|---|---|
| ファ2 | fa | ファイル fairu | "file" |
| フィ12 | fi | フィン fin | "fin" |
| フェ1 | fe | フェリー ferī | "ferry" |
| フォ1 | fo | フォーク fōku | "fork" |
A direct quotation shows the range: "In loanwords, [ɸ] can occur before other vowels or before /j/. Examples include [ɸiɴ] (フィン, fin, 'fin'), [ɸeɾiː] (フェリー, ferī, 'ferry'), [ɸaɴ] (ファン, fan, 'fan'), [ɸoːmɯ] (フォーム, fōmu, 'form')."1
These imports promote [ɸ] from a positional allophone to a near-phoneme, because they introduce minimal pairs that native vocabulary lacks. In a minimal pair, two words differ by only one sound. "[ɸoːkɯ] (フォーク, fōku) and [hoːkɯ] (ホーク, hōku) from English fork and hawk" are now distinguished only by the consonant.1
The extended-katakana series ファ・フィ・フェ・フォ is the standard way to write new loanwords. Older imports preserve the earlier repair, where foreign /f/ was reanalyzed as /h/ before non-/u/ vowels. That is why コーヒー kōhī ("coffee") begins with [k] and [oː] followed by [h], not [ɸ].12
Good to know
The candle-blow mnemonic
Blow out an imaginary birthday candle gently, freeze the lip shape, then voice the vowel into the channel between the lips.10 The drill works because it separates the bilabial gesture from the vowel and keeps the upper teeth out of the way. That is exactly the gesture [ɸ] needs.39 It is the most-repeated cue in English-language teaching materials on the Japanese fu sound.910
Why dictionaries still list both "fu" and "hu"
Hepburn (1867) writes ふ as "fu" to match the surface [ɸ] articulation. Kunrei-shiki (Japanese Cabinet Order No. 1 of 29 December 1954, ISO 3602:1989) writes the row systematically as ha-hi-hu-he-ho by kana position rather than surface sound.256 The 16 December 2025 cabinet notification reversed Japan's official preference and made Hepburn the new national standard. Kunrei spellings still appear on older signage, in older linguistics papers, and in some Japan-internal documents.56 The sound is [ɸɯ̟ᵝ] either way. The spelling is a notation choice.
Treating ファ・フィ・フェ・フォ as English F
A learner reading ファイル fairu ("file"), フィルム firumu ("film"), or フォーク fōku ("fork") may default to English labiodental [f] because the romanization shows "f". The correct production keeps the same bilabial [ɸ] used in native ふ.12 The extended-katakana series exists so Japanese can write the new contrast that loanwords created, including the minimal pair "[ɸoːkɯ] (フォーク, fōku) and [hoːkɯ] (ホーク, hōku) from English fork and hawk." The consonant is still bilabial [ɸ], not labiodental [f].1
Producing English /f/ gives foreign-accented Japanese, not a more "accurate" loanword. The bilabial cue from native ふ transfers unchanged into every ファ・フィ・フェ・フォ environment.12
Reading ふ as a glottal "hu"
A learner who has been told "ふ is not English F" sometimes drops all lip activity and produces a fully glottal [h], so 服 fuku comes out as English "hoo-koo".
服を買いました。4
"I bought clothes."
The target is not the absence of friction. It is the relocation of friction to the lips: a clear bilabial [ɸ] onset, then the compressed close-back [ɯ̟ᵝ].129
Cross-language false friends
Spanish, German, and French /f/ are all labiodental, the same place of articulation as English /f/. Transfer from any of those L1s therefore lands in the same wrong place as English-speaker transfer.3 The candle-blow correction applies unchanged across these L1s, because the target articulation is L1-independent.10
Speakers who front [ɸ] toward [f]
Some speakers, in loanword-heavy registers, can drift toward a labiodental [f] in ファ・フィ・フェ・フォ environments. Aim for the bilabial [ɸ] across the board. The variation will not get marked as wrong, while a consistent labiodental will be heard as foreign-accented.
See also
- The Japanese ら-Row (ら り る れ ろ): The Alveolar Tap That Is Neither R Nor L
- Why "Tokyo" Is Two Syllables in English and Four Morae in Japanese: Loanwords as a Timing Drill
- Japanese Pronunciation Drills: A Daily 5-Minute Protocol with Minimal Pairs, Shadowing, and Record-and-Compare
- Common Romaji Mistakes That Mislead Pronunciation