9.4 — Where Words Come From

English has a vocabulary assembled by invasion, trade and empire, spelled according to how it was pronounced six hundred years ago, and it is still absorbing.

The shape of English

Why is English spelling so bad?

Because the spelling was fixed by printing while the pronunciation kept moving, and because the words come from too many places.

The Great Vowel Shift, running roughly from 1400 to 1700, changed how English long vowels were pronounced across the whole language. Bite had been closer to "beet", meet closer to "mate", house closer to "hoose". Printing arrived in the middle of it, in 1476, and printed spellings began to standardise while the sounds carried on shifting underneath.

So words are largely spelled as they sounded before the shift. The silent letters are frequently not silent historically: the k in knight and knee was pronounced, and so was the gh, as a sound like the ch in loch.

Two further complications. Scribes and printers added letters on purpose. The b in debt and doubt was inserted by scholars who wanted the words to look more like their Latin ancestors debitum and dubitare, although the words came into English from French without it. The s in island is the same kind of well-meaning error — the word is Germanic and has nothing to do with Latin insula.

And early printers included Flemish speakers, brought in because printing developed on the continent, who introduced spellings from their own conventions. The h in ghost is generally attributed to them.

The result is a spelling system that is a historical archive. It preserves information — you can often see a word's origin in its shape — at the cost of being learnable only by memorisation.

Where do English words actually come from?

In four large waves, each identifiable in the modern vocabulary.

Germanic, from the Angles, Saxons and Jutes from the fifth century. This is the core of the language: the grammar, the pronouns, the commonest words. House, water, food, eat, drink, sleep, man, woman, child, good, bad. Around a hundred of the hundred most-used English words are from this layer.

Old Norse, from Viking settlement in the ninth and tenth centuries. Sky, egg, knife, husband, window, leg, skin, anger, die — and, unusually, the pronouns they, them, their, since core grammatical words are rarely borrowed. That they were is evidence of how thoroughly the two populations mixed.

French, from the Norman conquest of 1066, which put a French-speaking ruling class over an English-speaking population for three centuries. The vocabulary of government, law, war, art and fine dining is French: parliament, judge, jury, court, army, soldier, art, beauty, cuisine.

Latin and Greek, borrowed continuously by scholars for scientific, medical and abstract vocabulary.

Why do we eat beef but farm cows?

This is the single clearest fossil of the Norman conquest in the language, and it is genuinely what it looks like.

The animal in the field was tended by English-speaking peasants: cow, pig, sheep, calf, deer — all Germanic. The meat on the table was eaten by French-speaking nobility: beef (bœuf), pork (porc), mutton (mouton), veal (veau), venison (venaison).

The pattern holds because the two groups encountered the animal at different points, and each used their own word for the thing they dealt with.

The same social split runs right through the vocabulary and produces the register difference that English writers exploit constantly. The Germanic word is plain and the French or Latin word is formal: ask/enquire, begin/commence, help/assist, end/terminate, freedom/liberty, kingly/royal/regal — that last set giving Germanic, French and Latin versions of one idea, at three levels of formality.

Which is why the advice to prefer short words has a historical basis. The short words are usually the old ones, which arrived first and carry more emotional weight.

Where words come from

Which English words come from Indian languages?

A large number, mostly through trade and administration, and most speakers do not recognise them.

From Hindi and Urdu: shampoo (from chāmpo, to press or massage), pyjamas, bungalow (from bangla, a Bengal-style house), jungle, loot, thug, dungarees, khaki (dust-coloured), chutney, cot, bangle, guru, pundit, cushy (from khush, happy), juggernaut (from Jagannath, per 5.5).

From Tamil: catamaran (kattumaram, tied wood), mango (via Portuguese from mangai), curry (from kari), pariah, cheroot.

From Sanskrit: avatar, karma, nirvana, yoga, mantra, sugar (via Persian and Arabic from sharkara), orange (via Persian and Arabic from naranga), candy (from khanda).

From Malayalam: teak, copra, coir.

From Bengali: dinghy, jute.

And veranda came into English from Hindi, which had it from Portuguese — which may itself have taken it from an Indian source, so the word may have made a round trip.

What is a word that means its own opposite?

A contronym, or Janus word, and English has a surprising number.

Cleave means to split and to cling to. Sanction means to permit and to penalise. Dust means to remove dust and to apply it. Oversight means supervision and failure to notice. Left means departed and remaining. Fast means moving quickly and firmly fixed — as in "hold fast", which is the older sense, and which is also why a fast dye is one that does not run.

They usually arise in one of two ways. Either two separate words converged in form — cleave is genuinely two different Old English verbs — or one word developed in two directions from an ambiguous core, as sanction did from a Latin word meaning to make binding, which can be done to a permission or a punishment.

Which words were invented by one identifiable person?

More than you would expect, and the identifiable ones are usually from writers.

Shakespeare is over-credited, per 9.1, but assassination and bedazzled are reasonably attributed.

Serendipity was coined by Horace Walpole in a 1754 letter, from a Persian fairy tale about the three princes of Serendip — an old name for Sri Lanka — who kept making discoveries by accident.

Robot comes from Karel Čapek's 1920 play R.U.R., from Czech robota, forced labour. Čapek credited the word to his brother Josef, who suggested it when Karel could not think of one.

Nerd first appears in Dr Seuss's If I Ran the Zoo (1950), as the name of an imaginary animal, and the connection to the later meaning is unproven but suspicious.

Quiz, according to a well-known story, was invented for a bet by a Dublin theatre manager who had the nonsense word chalked on walls overnight to see whether it would enter the language. The story is almost certainly false and appears long after the word is first attested.

Meme was coined by Richard Dawkins in The Selfish Gene (1976), from Greek mimeme, for a unit of cultural transmission that spreads by copying — which is exactly what happened to the word.

What is a folk etymology?

A false origin story that is more satisfying than the real one, and which sometimes changes the word to match itself.

"Posh" does not stand for "port out, starboard home". There is no documentary evidence for the shipping-ticket story, and the word appears in slang before the supposed practice. "Golf" does not stand for anything. Acronym-based etymologies are almost always false for any word predating the twentieth century, because acronyms as a word-forming device are essentially modern.

The genuinely interesting cases are where the false etymology changed the word. Crayfish has nothing to do with fish — it is from Old French crevice — but English speakers heard fish and respelled it. Hangnail is from Old English angnaegl, a painful spike, reinterpreted as a nail that hangs. Bridegroom is from guma, an obsolete word for man, reinterpreted as groom.

Sneezing. The English word derives from an older form fnese, and the change to sn- may be a misreading of the long s in manuscript, which looks like an f.

Language change

Why do languages change at all, and can it be stopped?

Because every generation learns the language from imperfect examples and reanalyses it slightly, and because language is used to signal group membership.

Change happens in several regular ways. Sounds shift systematically. Grammar simplifies where redundancy exists — English lost its case endings and gender because word order took over the work. Meanings drift: nice meant foolish, then precise, then pleasant; awful meant inspiring awe; decimate meant kill one in ten. And words are borrowed constantly.

Stopping it has been tried repeatedly and has never worked. The Académie Française has been ruling on correct French since 1635 and French has changed anyway. Prescriptive rules that people actually follow are the ones taught in schools and enforced socially, and even those decay.

Several famous English rules were invented rather than observed. The prohibition on splitting an infinitive was imported from Latin, where an infinitive is one word and cannot be split. The prohibition on ending a sentence with a preposition has the same origin. Neither describes how English has ever been used.

What is the difference between a language and a dialect?

Politically decided, and there is no linguistic test that works.

The intuitive criterion is mutual intelligibility: two speakers who can understand each other are speaking the same language. It fails immediately in both directions.

Hindi and Urdu are mutually intelligible in ordinary speech, use different scripts and different formal vocabulary, and are counted as separate languages. Serbian and Croatian are the same case. Danish, Swedish and Norwegian are largely mutually intelligible and are three languages with three states behind them.

The varieties of Chinese go the other way. Mandarin and Cantonese are not mutually intelligible at all — a Mandarin speaker cannot understand spoken Cantonese — and are conventionally called dialects of one language, because they share a writing system and a state.

The linguist Max Weinreich's line is the standard summary: a language is a dialect with an army and a navy.

How many languages are there, and how fast are they disappearing?

Around 7,000, and the count depends on exactly the question above.

The distribution is extremely skewed. About half the world's population speaks one of roughly 25 languages. The bottom half of the list — around 3,500 languages — is spoken by well under one per cent of people. Papua New Guinea alone has over 800.

A language is generally considered endangered when children stop learning it, because at that point the remaining speakers are a fixed and shrinking set. By that measure, something in the region of 40 per cent of the world's languages are endangered, and estimates commonly suggest one dies roughly every two weeks.

What is lost is not sentimental. A language encodes classifications its speakers found worth making — botanical and medical knowledge, navigation, kinship, environmental detail — and much of it exists nowhere else. Documentation work is a race, and revival is possible but hard: Hebrew is the only substantial case of a language with no native speakers becoming a national language with millions, and it took a state, a school system and a century.

What comes next

The last page of this Part is about the systems used to write all of this down — alphabets, syllabaries and the scripts of the world, including the several that have never been read.