British, American and Indian English: spelling, vocabulary and register

What this corpus can and cannot tell you

Start with the limits, because they shape everything below. This dictionary marks a sense as regional only when the meaning itself is regional — 918 entries carry a United Kingdom usage label, 126 an American one, 82 Indian or subcontinental. Those are small numbers against 147,478 entries, and deliberately so.

It does not mark spelling variants as regional at all, because to a lexical database “colour” and “color” are the same word with the same meaning. Both are here, both have full entries, and neither is flagged.

There is one measurable side effect of that worth knowing about. Our commonness bands are computed from a frequency corpus dominated by American-spelled text, so in every variant pair the American spelling sits a band higher: color is Core where colour is Common, center is Core where centre is Common, defense is Core where defence is Common. That is a fact about the measuring corpus, not about the words, and you should read the bands on those pairs accordingly.

The spelling patterns, in order of how often they bite

Almost all the spelling difference you will ever meet is these six patterns. Indian English follows the British column throughout.

The -ise / -ize rule almost everyone gets wrong

“-ize is American” is the most repeated and least accurate of the spelling rules. -ize is the older form in English, it is the house style of Oxford University Press, and it is used throughout British academic and scientific publishing. Both organise and organize are British; only organize is American.

There is a genuine catch. A group of verbs are never spelled -ize in any variety, because their -ise is not the suffix at all: advertise, advise, comprise, compromise, despise, devise, exercise, improvise, revise, supervise, surprise, televise. Write “advertize” anywhere and it is simply wrong.

So the safe rule is the opposite of the one people repeat: -ise is safe everywhere, -ize is safe in America and in British academic writing, and there is a fixed list of verbs where only -ise is possible.

Vocabulary: fewer differences than the lists suggest

The famous pairs — lift and elevator, flat and apartment, pavement and sidewalk, biscuit and cookie, boot and trunk — are all real, all in this dictionary, and all mutually understood. Neither side of any of them needs translating; they are marked, not opaque.

The ones that cause actual trouble are the false friends, where the same word means different things. Ask for a “rubber” in an American office, or say a proposal has been “tabled” — in British use tabling something puts it on the agenda, in American use it takes it off. Nothing sounds foreign and the meaning inverts.

Indian English adds vocabulary rather than replacing it. Lakh and crore for hundred-thousand and ten-million are in this dictionary as ordinary entries; so are ashram, dacoit, godown, guru, karma, mantra and caste. They are not informal and they are not slang — in Indian newspapers and financial reporting, crore is the standard unit.

The gap our own data has here

Both lakh and crore are in this corpus with definitions, and neither has a pronunciation. The reason is exactly the kind of thing worth surfacing rather than hiding: our transcriptions come from an American pronouncing dictionary, and words standard to several hundred million English speakers are simply not in it.

That is not a small caveat. 35,490 of 83,117 single words have a transcription; the missing ones are not randomly distributed, and vocabulary from outside British and American English is over-represented among them.

The definitions have their own centre of gravity too. WordNet was built in the United States, so its coverage of American vocabulary and American senses is denser than its coverage of Indian, Nigerian, Singaporean or Australian English. Where a dictionary's sources come from shows up in what it happens to know.

Register is the difference that actually matters

Spelling is a surface you can convert mechanically. Register — how formal a word is, and whether it is the expected word in a given kind of writing — does not convert, and it is where writing sounds foreign long after the spelling is consistent.

This dictionary labels register where WordNet does: 610 entries are marked informal, 202 slang, 130 archaic, 208 trademarks. Those labels are shown on the sense that carries them. What no label can tell you is that a word neutral in one variety reads as stiff or as casual in another, because that is a fact about a readership rather than about a word.

The one piece of advice that holds regardless of variety: pick one and be consistent within a document. Readers barely register which convention you chose. They notice immediately when it changes halfway down the page.

Questions

Is -ize American and -ise British?

No. -ize is standard in American English and also in British academic publishing, including Oxford University Press house style. -ise is standard in most other British and Indian writing. A fixed group of verbs — advertise, advise, surprise, televise and about a dozen more — are never -ize anywhere.

Does Indian English use British or American spelling?

British, consistently: colour, centre, organise, travelled. The vocabulary adds terms of its own — lakh, crore, ashram, godown — which are standard rather than informal in Indian publishing.

Which spelling does this dictionary use?

Both. Every variant pair has two full entries with the same definitions. What you should know is that the commonness band is measured against a mostly American-spelled corpus, so the American spelling of a pair reads one band commoner than its British twin.

Which accent do the pronunciations show?

General American, from the CMU Pronouncing Dictionary. That also means vocabulary from outside American English is over-represented among the 47,627 single words that have no transcription here at all — lakh and crore among them.