SpellingIQ

Where English spelling comes from

English is several spelling systems at once, one for each language it borrowed from. Knowing which one a word belongs to explains most of what looks arbitrary.

One language, several spelling systems

Linguists who have surveyed English spelling agree that it is not one system but several, sorted by the route a word took into the language. K. H. Albrow first described it that way in 1972, a native core beside a Latinate layer and a residue of recent loans; Edward Carney's survey files every sound-to-letter correspondence into subsystems by origin and finds them far more regular inside a subsystem than across them (Albrow, 1972; Carney, 1994). Richard Venezky's structural account of American English reaches the same conclusion (Venezky, 1999).

The practical consequence is that a spelling which looks arbitrary on its own usually looks regular once you know which layer it belongs to. Ph for the f sound is odd in English and perfectly ordinary in a Greek word; a silent final t is odd in English and ordinary in a late French loan. This is also part of why the classic computational analysis of 17,310 words found the share of English that is genuinely unpredictable falling to a few percent once origin and meaning were counted (Hanna et al., 1966), a figure that is quoted more often than re-derived and is best read as an order of magnitude. The layers below are the ones that account for most of the corpus of everyday and competition words; each note says where its history is told.

Native English

The oldest layer. Old English was spelled as it sounded, and the sounds moved: the k of knight, the w of write and the gh of night were all once pronounced, and the letters stayed after the sounds went. Middle English scribes added tch and dge for the ch and j sounds after a short vowel.

  • ight for the ite sound · night, knight
  • ough, one spelling for several sounds · though, through
  • silent gh · weigh, taught
  • silent k before n · knife, know
  • silent w before r · write, wrist
  • wh for the w sound · which, whisper
  • silent g beside n · gnaw, gnat
  • silent b after m · thumb, lamb
  • tch for the ch sound · kitchen, watch
  • dge for the j sound at the end · bridge, edge

Sources: Crystal (2012); Horobin (2013); Carney (1994).

French

The largest borrowed layer, in two waves. After 1066 French scribes respelled English with their own habits (qu, ou, gn), and French words came in by the thousand. Later loans kept French endings as they were: que and gue for the k and g sounds, ette, eur, eau, a silent final t or s.

  • eau for the o sound · bureau, plateau
  • que for the k sound at the end · boutique, technique
  • gue for the g sound at the end · fatigue, league
  • ette at the end · silhouette, cigarette
  • eur at the end · amateur, entrepreneur
  • silent u after g before e or i · guide, guitar
  • silent g beside n · campaign, foreign
  • ch for the sh sound · chef, machine
  • oi for the wa sound · reservoir, memoir
  • ou for the oo sound · soup, route
  • et for the ay sound at the end · ballet, gourmet
  • silent final t · debut, depot
  • silent final s · debris, chassis
  • qu for the k sound · masquerade, conquer

Sources: Crystal (2012); Carney (1994); Venezky (1999).

Latin

Words taken straight from Latin, mostly from the Renaissance on, keep Latin's letters: tion, ous, the sc of science, ae and oe. Scholars also respelled older words to look Latin, which is where the b of debt and doubt comes from.

  • tion for the shun sound · nation, station
  • ous for the us sound · famous, nervous
  • silent c after s before e or i · science, fascinate
  • ae for the e sound · minutiae, formulae
  • oe for the e sound · subpoena

Sources: Crystal (2012); Carney (1994); Venezky (1999).

Greek

Greek words reached English through Latin, and Latin wrote Greek letters its own way: ph for phi, ch for chi, y for upsilon, rh for rho with its breathing, ps and pn for letter pairs English cannot begin a word with. The spelling records the Greek letter, not the English sound.

  • ph for the f sound · phenomenon, metaphor
  • ch for the k sound · chasm, chronicle
  • y for the i sound between consonants · cylinder, syllogism
  • rh for the r sound · rhetoric, rhinoceros
  • silent p before s · psalm, psychology
  • silent p before n · pneumonia, pneumatic
  • silent p before t · pterodactyl, ptomaine
  • silent m before n · mnemonic
  • x for the z sound at the start · xylophone, xenophobia
  • ae for the e sound · archaeology, daemon
  • oe for the e sound · phoenix, amoeba

Sources: Crystal (2012); Carney (1994).

Italian

Italian loans keep Italian spelling, which is regular by Italian rules: cc before e or i is the ch sound, cch and ch are k, gn is ny, zz is ts, sch is sk.

  • ch for the k sound · chiaroscuro, archipelago
  • cch for the k sound · zucchini, gnocchi
  • sch for the sk sound · bruschetta, scherzo
  • cc for the ch sound before e or i · cappuccino, focaccia
  • zz for the ts sound · pizza, paparazzi
  • gn for the ny sound · lasagna, gnocchi

Sources: Carney (1994); Venezky (1999).

Spanish

Spanish loans keep Spanish spelling: ll for the y sound, j for the h sound, qu for k before e and i. English often keeps the letters and changes the sound, as in vanilla and guerrilla.

  • ll for the y sound · tortilla, quesadilla
  • j for the h sound · jalapeño, fajita
  • qu for the k sound · quesadilla, tequila

Sources: Carney (1994); Venezky (1999).

German

German loans keep German spelling: sch for the sh sound, tz and z for ts, w for v, au for ow. The later the loan, the more of the German sound English keeps with the letters.

  • sch for the sh sound · schadenfreude, schnapps
  • tz for the ts sound · ersatz, pretzel
  • w for the v sound · weltschmerz, zugzwang
  • z for the ts sound · zeitgeist, ersatz
  • au for the ow sound · sauerkraut, umlaut

Sources: Carney (1994); Venezky (1999).

Arabic

Arabic loans arrive by transliteration, and the conventions for writing Arabic in Roman letters allow q without a following u, as in Qatar and Iraq, and use kh for a sound English has no letter for, as in sheikh.

  • q without u for the k sound · qatar, iraq
  • kh for the k sound · sheikh

Sources: Carney (1994).

Japanese

Japanese loans are written by the standard romanisation, which allows ts at the start of a word where native English spelling does not.

  • ts at the start · tsunami

Sources: Carney (1994).

Where the layers blur

The boundaries are analytic, not sharp. A French word taken early follows native habits and one taken late follows French: chance and chef both came from French, and only the second keeps the French sound of ch. Choice and reservoir share an oi that reads two ways. English also reshaped many loans to its own pronunciation while keeping the letters, which is why vanilla and guerrilla no longer sound their Spanish ll. So the conventions above describe what a family's letters usually mean, not a rule with no exceptions, and the origin that matters is the language a word arrived from, not the oldest ancestor a dictionary can trace.

A caution about what this does and does not show. That English spelling is layered by origin is descriptive scholarship, well established and not in dispute. Whether being told a word's origin makes anyone a better speller is a different question, and one the research has not answered; the strongest evidence for teaching the structure of words concerns their meaningful parts rather than their history (see how people learn to spell). Competitive spellers organise their preparation around language of origin all the same, which says the cue is usable, not that it has been measured.

References