Phonological History and Comparative Linguistics
Why is fu the only odd one out? To answer that question, you need to know 1,300 years of the history of Japanese sound.
Anyone studying Japanese is bound to notice a certain imbalance in the ha-row. The kana は・ひ・ふ・へ・ほ all belong to the same row, yet when you romanize them you get ha, hi, fu, he, ho — and fu alone is spelled with an f. In phonetic terms, fu is a bilabial fricative [ɸ]: you bring your upper and lower lips slightly together and let air friction through the gap. By contrast, ha, he, and ho are glottal fricatives [h], produced at the back of the throat. And here’s a point that often goes unnoticed: hi isn’t a pure [h] either — it’s actually a third sound entirely, a palatal fricative [ç]. In other words, the five sounds of the modern ha-row have already split into three distinct places of articulation.
Linguistic irregularity is almost always a geological layer of history. What looks like “disorder” or an “exception” when you view it synchronically — as a snapshot of the language today — reveals a perfectly coherent logic of change when you lay it out diachronically, across time.
The Three Stages of Sound Change
| Period | Reconstructed sound | Articulation | “Flower” (hana) |
| Up to the Nara period (–8th c.) | *p (bilabial stop) | Lips closed, then released in a burst | pana |
| Heian to Muromachi (9th–16th c.) | ɸ (bilabial fricative) | Lips loose, air pushed through | fana |
| Edo period onward (17th c.–) | h (glottal fricative); ɸ in fu, ç in hi survive | Air friction at the back of the throat | hana |
The sequence that comparative linguistics has reconstructed goes like this. The ha-row was originally a bilabial stop, *p, and 花 (“flower”) was pronounced pana. Over time, the tension holding the lips closed relaxed, and the sharp stop softened into friction — the sound shifted to a bilabial fricative ɸ (fa, fi, fu, fe, fo). That relaxation kept going: the point of articulation retreated further back, from the lips all the way to the glottis, and the sound settled as h. The forward closure at the lips “pulled back” through friction until it came to rest at the furthest-back point possible — the throat. Historical phonology calls this process labial weakening (唇音退化, shin’on taika).
One important caveat: the dates in the table are guideposts for the stages, not hard boundaries. The shift from P to F is thought to have been well underway during the Heian period — considerably earlier than the Muromachi-period documents that give us our clearest written evidence. (The phenomenon known as ha-gyō tenkoion, discussed below, is one of the markers of that earlier transition.) The final handover from F to H is the relatively recent change that happened in the Edo period.
Fu and Hi — Two Sounds Left Behind
The retreat from ɸ to h didn’t proceed at the same pace across all vowel combinations. Here’s how each played out:
- ɸa → ha (は)
- ɸi → çi (ひ) — drawn forward by the front high vowel /i/, it underwent palatalization and settled as [ç]
- ɸu → ɸu (ふ) — due to its compatibility with the vowel /u/, ɸ simply stayed put
- ɸe → he (へ)
- ɸo → ho (ほ)
The key to fu holding on as a bilabial fricative [ɸ] lies in its vowel, /u/. Japanese /u/ isn’t rounded in the typical sense — it’s a compressed vowel [ɯᵝ], where the lips are narrowed from the sides rather than pushed forward. That lip posture overlaps with the posture required for the bilabial fricative ɸ. Because the consonant and vowel already share the same lip movement, there was simply no pressure pushing the articulation further back toward the throat. In a sense, fu was sheltered by its own vowel, which allowed it to preserve the older form. Hi underwent its own version of this: drawn toward the vowel /i/, it branched off into the distinct sound [ç] rather than merging into the glottal [h]. Fu tends to get all the attention, but strictly speaking, hi is equally an “exception” that failed to complete the ha-row’s shift to H.
The Evidence — What Documents and Structure Tell Us
That the ha-row was once a P and F sound is not speculation. It is confirmed by multiple independent lines of evidence.
Ha-gyō tenkoion and the particles は and へ During the Heian period, ha-row sounds in the middle or at the end of words shifted to [w] while retaining their labial character — a phenomenon called ha-gyō tenkōion (“labial turn”). This is why かは (kawa, “river”) became かわ, and こひ (kohi, “love”) became こい. The modern spelling 川 = かわ is a direct result of this change. What’s especially telling is the convention of reading the particle は as wa and へ as e. This is a fossil frozen in the writing system: the labial shift of Heian Japanese that happened in word-medial position got locked into the orthography, and so every day, without thinking about it, we’re pronouncing traces of Heian Japanese in our own sentences.
Missionaries recording the F sound — the Nippo Jisho In the early 17th century, Jesuit missionaries who had come to Japan recorded the Japanese language in Portuguese-style romanization. In the Nippo Jisho (1603–04), the ha-row is consistently spelled with f throughout. “Flower” is fana, “person” is fito, “Japan” is Nifon. The living pronunciation of Japanese at that moment was captured objectively through foreign ears and an alphabet — making this direct evidence that Japanese was still in the F-sound stage in the early Edo period.
The Gokashiwabara-in riddle In a Muromachi-period riddle collection, Gonara-in Gosen Nanzo (ca. 1516), there is a famous puzzle: “I have met my mother twice, but my father not even once.” The answer is “lips” (kuchibiru). Pronounce haha (“mother”) and your lips touch twice; pronounce chichi (“father”) and they never touch at all. The riddle only works because “mother” was pronounced ɸaɸa — “fafa” — in the speech of the time. With the modern pronunciation haha, the lips never meet, and the puzzle falls apart entirely.
P sounds fossilized in gemination and compound words The even older *p lives on, trapped inside the geminate consonants (the small っ) of compound words. 一 (ichi) + 本 (hon) = ippon; 十 (jū) + 本 = juppon; 切 + 符 = kippu. The strong phonological environment of gemination blocked the sound from undergoing fricativization, preserving the oldest sound value intact. The ha-row reads as h in isolation, yet suddenly p surfaces in these environments — like the original stratum breaking through to the surface from deep underground.
Evidence in Modern Dialects — Ryukyuan Languages as a “Museum of Time”
The sound changes of central Japanese did not spread uniformly across every corner of the archipelago. The southern dialects in particular have preserved older stages frozen in place.
- Pana (Okinawa, Amami, and neighboring areas) — the oldest transmitted form of “flower.” This preserves the P-sound stage that the central language left behind long ago, and shows that the Ryukyuan languages, after branching from the same common ancestor as Japanese, maintained the ancient form on their own.
- Fana (parts of Okinawa) — “flower” halted at the F-sound stage. Remarkably, this is the same phonological system as the Muromachi-period central language — or the fana of the Nippo Jisho.
The three-stage journey — pana → fana → hana — that the central language ran through and moved on from has been “frozen” at various points around the archipelago according to geographical distance from the center. The spread of space translates directly into depth of time.
Linguistic Implications — Conservative Retention and “Living Fossils”
Language change is never uniform. Even as the whole system moves toward a new stage, the older form stubbornly survives in particular phonetic environments (ɸ followed by /u/), particular structures (geminates, compound words), or regions far from the center. This is what linguists call conservative retention.
The existence of fu is unshakeable evidence that the ha-row once belonged to an F-sound system. The p in ippon tells us that an even older P-sound era lies beneath that. And the diversity of the Ryukyuan languages maps the entire timeline of these changes not as a flat chronology but as a three-dimensional landscape.A small sense of something off — felt when looking at a single kana — threads together three independent witnesses: written documents, linguistic structure, and living dialects, and opens into a dissection of 1,300 years of sound history. Fu is a living fossil in the ecosystem of Japanese: proof that the past is still breathing in the present.