The Graveyard of /p/

A record of where the voiceless bilabial plosive went to die.

However language began, one thing is certain — it immediately began to change, and has been changing ever since.

David Crystal · 2007

時有古今。
地有南北。
字有更革。
音有轉移。

陳第 · 1606

What is /p/?

/p/ — the voiceless bilabial plosive — is one of the most widespread phonemes in human language. And yet it is absent from roughly 10% of the world's languages, and in many others it has weakened, shifted, or vanished under specific phonological conditions.

This site records those losses. The diagram below is a phonological flow chart — each ribbon is one lineage of /p/ over time. A ribbon's color is the phoneme as it's articulated at that year (amber for /p/, blue for /f/, violet for /ɸ/, green for /h/, teal for /ç/, rose for /w/, grey for ∅). When a single /p/ splits into multiple environments, ribbons fork. When the color shifts mid-flow, that's a sound change in progress. Pre-evidence sections fade in from transparent — we know /p/ was there, but its exact form is extrapolated. Dots are attestation points. The ribbons are grouped by language family so you can see, e.g., the fates of Indo-European *p side by side.

Drag horizontally to pan; use the zoom controls to expand a period. Hover any ribbon or dot for details.

Timeline

100%
/p/ ∅ /f/ /ɸ/ /h/ /ç/
3500 BCE3000 BCE2500 BCE2000 BCE1500 BCE1000 BCE500 BCE0500 CE1000 CE1500 CE2000 CEPIE *pCelticGermanicItalic / RomanceGreekIndo-IranianProto-Semitic *pArabicHebrewAkkadianOld Chinese *p重唇音輕唇音Proto-Japonic *pOld Japanese{Q,N}_VV_V / _{a,o,e}V_V / _iV_V / _uLoanwords3500 BCE — PIE *ph₂tḗr (reconstructed)500 BCE — Gaulish atir (/p/ already gone)800 CE — OIr. athair "father" (cf. Lat. pater)200 BCE — Grimm's Law (PIE *p → PGmc *f)800 CE — OE fæder (< PGmc *fader)700 BCE — Old Latin (/p/ preserved)1300 CE — Romance daughters (/p/ kept across the board)1450 BCE — Linear B pa-te-re (Mycenaean)700 BCE — Homeric πατήρ1500 BCE — Vedic pitár3500 BCE — PSem *p (reconstructed)200 BCE — Old Arabic inscr.700 CE — Classical Arabic (no /p/ in inherited words)800 BCE — Biblical Hebrew ([p]~[f] allophony)2300 BCE — Old Akkadian (/p/ preserved)1000 BCE — OBI / Bronzeware (OC *p reconstructed)600 CE — Qieyun 切韻 (/p/ in all environments)2025 CE — 北 běi (/p/ preserved)1100 CE — 韻鏡 Yùnjìng (輕唇音 split codified)1324 CE — 中原音韻 (/f/ established)2025 CE — 飛 fēi, 風 fēng (modern /f/ < MC *pjV)300 BCE — PJ *p (reconstructed)720 CE — Man'yōgana (/p/ likely [p])2025 CE — 切符, 散歩 (kippu, sanpo)1100 CE — Heian /ɸ/1603 CE — Nippo Jisho "f" (/ɸ/ still attested)1900 CE — modern /h/1543 CE — Portuguese: pan, tabako1900 CE — English: pen, computer
hover a ribbon or evidence dot for details

The Graveyard

Celtic

Status: Lost in Proto-Celtic

One of the showpiece sound losses of Indo-European studies. PIE *p disappeared early in the Celtic branch, before Proto-Celtic split into its daughters. The classic cognates expose it:

  • PIE: *pPCelt: ∅ e.g. *ph₂tḗr → OIr. athair
  • Latin piscis ~ Old Irish íasc "fish"

/p/ later re-entered Celtic from kw in the P-Celtic branch (Welsh pedwar "four" < PIE *kʷetwores), but inherited PIE *p never came back.

Arabic

Status: Lost

Proto-Semitic had *p, retained in many sister languages — Hebrew shows it as [p]~[f] allophony, Akkadian had /p/. Arabic merged it with /f/ before the historical record:

  • PSem: *pArabic: f
  • Modern loans: park → bark (بارك) or fark (فارك)

Chinese

Status: Partially alive

Old Chinese had *p. In late Middle Chinese the labial initials split — 輕唇音 (light labials) and 重唇音 (heavy labials) — and *p underwent labiodentalization before specific medial-vowel combinations:

  • MC: pMand.: p elsewhere e.g. 北 běi
  • MC: pMand.: f _jɨ, _ju, _jo e.g. 飛 fēi, 風 fēng

Modern Mandarin keeps /p/ in plenty of words, but a large slice of the inherited vocabulary exited via /f/.

Japanese

Status: Resurrected

The most thoroughly documented case. Old Japanese /p/ weakened to /ɸ/ across most positions by the late Heian period, then split further by the following vowel:

  • OJ: pMJ: ɸ V_V
    • MJ: ɸmJ: h _{a, o, e}
    • MJ: ɸmJ: ç _i
    • MJ: ɸmJ: ɸ _u (marginal)
  • OJ: pmJ: p {Q, N}_V e.g. 切符 kippu, 散歩 sanpo

/p/ then re-entered the language wholesale: first via Portuguese loans (1543: pan, tabako), then Dutch, and at scale through modern English loans.

References

Frellesvig, Bjarke (2010). A History of the Japanese Language. · Baxter & Sagart (2014). Old Chinese: A New Reconstruction. · Matasović (2009). Etymological Dictionary of Proto-Celtic. · Lipiński (2001). Semitic Languages: Outline of a Comparative Grammar.