Spelling reform

Why does English spelling stay irregular when almost everyone agrees it is a burden?

Through, though, thought, thorough, tough. Every literate English speaker has complained about this, and the complaint is not merely aesthetic: children learning English take visibly longer to reach fluent decoding than children learning shallower orthographies like Finnish or Italian, a gap the large cross-language comparisons have repeatedly found.

The usual explanation is inertia — nobody got round to it. That is testable, because inertia predicts that spelling reform is not the sort of thing that happens. It plainly is. Russia reformed its orthography in 1918, Turkey changed script entirely in 1928, Germany in 1901 and again in 1996, the Netherlands several times, Portugal and Brazil in 1911, 1943 and 1990, Indonesia in 1972. So the question is not why writing systems resist change. It is why this one does.

Ask what those cases had in common and one thing stands out: a state, or a state-backed academy, with authority over schooling, examinations and official printing. Somebody could change what children were taught and what the government printed, on a date, for everyone at once.

Why should that matter so much? Because the value of a spelling is entirely in being shared. Writing thru helps me only to the extent that others read it without friction. Someone who unilaterally adopts a better system does not get a better system — they get a worse one, because they are now harder to read. The payoff to moving first is negative and stays negative until enough others have moved. This is not inertia, which implies a stationary object needing a push. It is an equilibrium that actively punishes departure.

The way out of such a trap is either an authority that moves everyone at once or an obvious alternative everyone can guess the others will pick. English has neither. It has no academy and no owner; it is the official or dominant language of dozens of sovereign states, none of which can bind the others, and its publishing, examination boards and dictionaries are private and competing. There is no lever of the kind Ankara or Berlin pulled.

Look at the two times English came closest. Webster's American dictionaries, from 1828, did change spellings — color, center, traveled — but got them adopted in one country only. The result was not reform. It was a permanent fork: two standards where there had been one, so every professional writer must now know both and the burden went up. That is exactly what a coordination game predicts when a subset moves. Then in 1906 the Simplified Spelling Board persuaded Theodore Roosevelt to order the Government Printing Office to adopt three hundred simplified spellings; the House objected, the press mocked it, and he withdrew the order within months. A head of state without the schooling lever was not enough.

A second obstacle undercuts the word agrees in the question. Almost everyone agrees the spelling is a burden. Nobody agrees on the replacement. A phonemic spelling must be phonemic for some particular accent, and English vowel systems diverge sharply between Scotland, southern England, North America, India and Australia — transparent for one is opaque for the others, where German and Dutch reformers each had a single national standard to anchor to. So the reformers split into incompatible camps: Shaw's alphabet, the initial teaching alphabet, Cut Spelling, a dozen more, none focal. A coordination problem with one obvious alternative is hard; one with twenty is much harder, because even those who want to move cannot say where to.

And the incumbent is not purely arbitrary. English spelling is partly morphophonemic: it holds a morpheme's shape constant across sound change, so sign and signature, heal and health stay visibly related, and the past tense is written -ed whether it sounds like t, d or id. Chomsky and Halle's claim that the system is therefore "near optimal" is contested and generally judged overstated — it discounts the real cost to beginners — but the trade-off it points at is genuine.

Finally, notice who pays. The cost falls on every literate adult, whose reading skill is capital denominated in the old spelling, and on publishers, libraries and teachers. The gain accrues to people not yet literate, who cannot advocate for themselves.

Two clarifications keep this from becoming a story about stubbornness.

The mechanism is not that people love the spelling. Its benefit is a shared convention, so individual and collective incentives point opposite ways and no actor has standing to resolve the conflict. That is the same shape as agreeing a plug or a tuning pitch, with one distinction worth naming: those were settled by bodies created for the purpose before the installed base grew large, whereas spelling's installed base is a billion literate people and every book ever printed. It also differs from an organisation keeping a rule everybody dislikes — there, somebody has the authority to repeal it and declines. Here nobody has the authority at all.

The second clarification is that "everyone agrees it is a burden" and "everyone agrees on the fix" are different propositions, and only the first is true. The absence of a focal alternative is not a footnote to the coordination problem. It is half of it.

• Why the 1996 German reform succeeded institutionally yet was partly rolled back after public resistance.
• Whether spell-checkers and predictive text have quietly reduced the burden that reform was meant to remove.

Every term the collection defines is gathered in the glossary.