Greek & Latin Roots of English

A word on a drug label

A cardboard box of glyceryl trinitrate carries an instruction that nobody explains to the patient: one tablet to be taken sublingually. Take the word apart and it stops being jargon. Sub is Latin for under. Lingua is Latin for tongue. The ending -al turns a noun into an adjective, and -ly turns an adjective into an adverb. Under-tongue-ly. Put the tablet under your tongue, where the mucous membrane is thin and the drug reaches the blood in about a minute instead of the half hour a swallowed tablet takes.

Nothing in that parse required Latin. It required knowing that the word has parts, that the parts sit in a fixed order, and that each part does one job. That is the whole method of this course, and this lesson sets it up.

The condition the tablet treats has a parseable name too. Angina is Latin, from angere, to squeeze or strangle, which is also the ancestor of anguish, anxious and anger. A patient with angina is being throttled from the inside, which is a better description of the symptom than most modern coinages manage.

Key idea: Roughly sixty per cent of the English vocabulary came from Latin and Greek, most of it in pieces that can be reassembled. A word from that layer is not a single lump to be memorised. It is an assembly, and assemblies come apart.

The four slots

A classical English word is built out of up to four kinds of part, in this order. Linguists call the parts morphemes, meaning the smallest pieces of a word that carry meaning of their own.

A word may have none of some slots and several of others. Antidisestablishmentarianism, the standard playground answer to the question of the longest word, is a prefix, a prefix, a base, and four suffixes stacked in sequence: anti plus dis plus establish plus -ment plus -arian plus -ism. Work outward from the base and it decodes without effort: establish, then the state of being established, then a person who supports the position, then the doctrine that person holds, then opposition to it. The word is long because it is precise, not because it is difficult.

Notice that the last piece of sublingually, the -ly, is not Latin at all. It is the native English adverb ending, from Old English -lice. English attaches its own grammar to borrowed material without a second thought, and that habit is the reason this course exists.

Why the base cannot stand alone

Here is the difficulty that makes classical vocabulary feel opaque. In a native compound such as bookshop or underhand, both parts are words you already know. In a classical word, the base is usually bound: it never appears on its own. Nobody ceives anything. There is no such verb as ceive, and yet English has receive, deceive, conceive, perceive and their whole family of nouns in -ception.

A bound base is not learned by looking it up in isolation. It is learned across a family. The most efficient way to see one is to lay the prefixes down one axis and the bases along the other.

Read one row across and the base holds still while the prefix does the work: to submit is to send yourself under, to remit is to send back, to emit is to send out. Read one column down and the prefix holds still while the base changes: sub- turns up under a tongue, under a duct, under an assumption. Seven bases and seven prefixes is forty-nine words, which is more than most people learn from a vocabulary list in a month.

The point: You do not memorise classical words one at a time. You memorise about two hundred bases and about forty prefixes, and the combinations come free.

The connecting vowel, in one paragraph

When two elements are joined and the join would be unpronounceable, a vowel goes in between. Latin uses -i-: agriculture is agr- plus -i- plus culture, carnivorous is carn- plus -i- plus vor- plus -ous, magnify is magn- plus -i- plus -fy. Greek uses -o-: thermometer, psychology, geography, chronology. That single vowel is one of the most reliable tells in the language for which of the two source languages an element came from, and Lesson 5 gives it a full treatment along with the cases where the vowel drops out.

Suffixes are part-of-speech machinery

A suffix rarely changes what a word is about. It changes what the word can do in a sentence. Learn the three families and you can convert freely.

Two of these carry meaning as well as grammar and are worth flagging now. The pair -er and -or names the agent, the one who does it; -ee names the one it is done to. An employer employs an employee; a lessor grants a lease to a lessee; a payer pays a payee. And -ist names a person by their doctrine or trade while -ism names the doctrine itself, which is why a socialist holds socialism and not the reverse.

Three parses, worked

Antipyretic. The label on a bottle of children's paracetamol may describe it as antipyretic. Anti- is Greek for against. The base is pyret-, from pyretos, fever, itself from pyr, fire, which also gave English pyre, pyrite and pyromania. The suffix -ic makes an adjective. Against fever. Now the next word on that label, analgesic, decodes on the same principle: an- for not, alg- from algos, pain, -esic making it an adjective. Not-pain.

Insectivorous. The base carries a small history lesson. Latin insectum means cut into, from in- plus secare, to cut, which is also behind section, dissect, segment and sector. It describes a body pinched into visible segments. The Latin word was itself a straight translation of the Greek entomon, also meaning cut up, which is why the study of insects is entomology rather than insectology. Add the connecting -i-, then vor- from vorare, to devour, then -ous. Insect-devouring.

Circumnavigate. Circum- is around. Nav- is navis, ship, as in naval and navy. Then something odd: -ig-, which is the Latin verb agere, to drive, with its vowel worn down inside a compound. To drive a ship around. That worn-down vowel is not an accident, and Lesson 9 is about the rule that produced it; for now, simply note that a base can arrive slightly dented and still be recognisable.

Worth holding on to: Work from the base outward, not from left to right. The base tells you what the word is about; everything else adjusts it.

Where the method stops

Not every English word yields to this. Outrage looks like out plus rage and is neither: it is Old French outrage, from outre, beyond, plus the noun-forming -age, and it entered English before anyone thought about rage. Handicap is hand in cap, from a wagering game. Belfry has no bell in it. The native layer of English, which supplies almost all the commonest words in the language, does not use classical parts at all: get, put, hand, home, work, ride, think, drink. When a parse feels forced, it usually is, and Lesson 10 is devoted to telling the difference. The method covers the borrowed classical layer, which is most of the technical vocabulary and most of the formal register, and it covers that layer very well.

Common misconceptions

• Long words are harder than short ones. Usually the reverse. Electroencephalography has one meaning and six visible parts. The word set has, by the count in most large dictionaries, well over a hundred senses and no parts at all.
• The base must be the longest chunk. In circumspect the base is four letters and the prefix is six. Size says nothing about which slot a piece is filling.
• Every English word can be parsed this way. Only the borrowed classical layer can. The native Germanic core, and words from Old Norse, Arabic, Dutch, Hindi and everywhere else, will not decompose into Latin parts, and forcing them produces folk etymology.
• The -ly on sublingually is Latin. It is native English, from Old English -lice. Mixed words like this are entirely normal and are the subject of Lesson 16.

Putting it together

• A classical English word occupies up to four slots in order: prefix, base, connecting vowel, suffix. Sublingually fills three of them and adds a native English adverb ending.
• Classical bases are usually bound, appearing only inside words, which is why they are learned across families rather than singly.
• A prefix-by-base grid generates whole word families at once and shows you which combinations the language actually took up.
• Suffixes mostly assign part of speech: -tion and -ment make nouns, -ous and -ic make adjectives, -ate and -ify make verbs.
• Latin joins with -i-, Greek joins with -o-, and that vowel is a first clue to which language you are dealing with.
• The method applies to the borrowed classical layer only. Native words, and words with a history the spelling has hidden, must be checked rather than parsed.

Sources

1. Wikipedia. (n.d.). Morpheme. Wikimedia Foundation. en.wikipedia.org
2. Wikipedia. (n.d.). List of Greek and Latin roots in English. Wikimedia Foundation. en.wikipedia.org
3. Harper, D. (n.d.). Outrage. Online Etymology Dictionary. etymonline.com
4. Harper, D. (n.d.). Insect. Online Etymology Dictionary. etymonline.com
5. Ayers, D. M. (1986). English words from Latin and Greek elements (2nd ed., revised by T. D. Worthen). University of Arizona Press.
6. Green, T. M. (2014). The Greek and Latin roots of English (5th ed.). Rowman and Littlefield.

Two words on the first page of an anatomy handout

A first-year student is told that afferent fibres carry signals toward the central nervous system and efferent fibres carry them away. She has the prefix list in front of her: ad- means toward, ex- means out of. Neither af- nor ef- is on it. The list is not wrong and neither is the handout. What is missing is a rule that everyone assumes you already know.

Say ad-ferre out loud at speed. The d, made with the tongue behind the teeth, is followed at once by f, made with lip against teeth. The tongue gives up, the d becomes f, and the word is affer-. Latin speakers did this constantly, Latin spelling eventually recorded it, and English borrowed the recorded form. That is assimilation, and it is the single largest cause of a prefix becoming invisible.

Key idea: The final consonant of a prefix tends to take on the character of the consonant that follows it. When you see a doubled consonant near the front of a Latinate word, your first hypothesis should be an assimilated prefix, not a strange base.

ad- in ten shapes

The prefix ad- means to, toward or in addition. It survives intact in adjacent, adhere and adverb, but in front of most consonants its d gives way.

Read the restored column and some quiet jokes come loose. To arrive is ad ripam, to come to the riverbank. An arrogant person claims things toward himself. To be aggravated is to have weight added.

in-, and the word that means the opposite of what it looks like

English inherited two separate Latin prefixes that both came out as in-, and both assimilate identically, which is the source of a great deal of confusion.

• Negative in-, the same element as English un- and Greek a-: il- before l (illegal, illiterate, illogical), im- before b, m and p (imbalance, immortal, impossible), ir- before r (irregular, irrational, irreverent), and i- before gn (ignoble, ignominy, and ignore, which is in- plus gnarus, not knowing).
• Directional in-, meaning in, into or on: illuminate (light shone into), immerse (dipped in), irrigate (water led in), invade, include, inscribe, irradiate.

Now the famous casualty. Inflammable does not mean fireproof. Its in- is the directional one: in plus flamma, to burst into flame. The word had meant exactly what flammable means for three hundred years before anyone worried about it, and the worry was justified, because people were reading the other in-. From the 1920s onward fire-safety bodies pushed flammable onto tankers and drums for that reason, and the labelling changed. The etymology never did.

The same trap sits inside invaluable, which means too valuable to be valued rather than worthless, and inhabitable, which means fit to live in, so that its opposite needs a second negative: uninhabitable.

The point: When you meet in-, decide which of the two prefixes you are looking at before you decide what the word means. The spelling will not tell you; the sense of the base usually will.

Four more that shift, and one pair that looks alike and is not

The trap in that table is sur-. Surrogate and surreptitious are sub- assimilated. Surface, surplus, surpass, surname, surtax and survive are not: their sur- is French for super-, meaning over or beyond. A surrogate stands under; a surplus sits over. Same three letters, opposite direction, and only the meaning of the word will tell you which you have.

Note also what ob- and sub- do before a consonant cluster: the prefix loses its own consonant rather than doubling. Omit is ob plus mittere; sustain and suspend are sub plus tenere and sub plus pendere.

The doubled consonant that is not a prefix at all

The rule cuts both ways, and over-applying it is how a confident parser gets things wrong. Some doubled consonants belong to the base and were doubled in Latin before any prefix arrived.

• annual, annals, anniversary, perennial, biennial. The base is annus, year, which already has two n. There is no ad- here.
• irritate, irritable. Latin irritare, to provoke. Not in- plus a base beginning with r.
• terrible, terror, deter. Latin terrere, to frighten. Not a prefix.
• collar, collateral in part, and colleague. Collar is collum, neck, whose double l is original; colleague is genuinely con- plus legare.
• summit, summary, consummate. Summus, highest, is itself a worn-down superlative and its double m is old.

The check is easy: if removing the supposed prefix leaves a base that means nothing, the doubling is probably original. Strip ann- from annual and you have nothing; strip the ad from accept and capere is waiting there.

Greek assimilates too, and it aspirates

The Greek prefix syn-, with, behaves exactly like Latin con-: sym- before b, m and p (symbol, symmetry, sympathy, symphony, symposium), syl- before l (syllable, syllogism, syllabus), sys- before s (system, systole), and sy- before z (syzygy, the alignment of three bodies in space). The prefix en-, in, does the same: emblem, emphasis, empathy, and ellipse, which is en plus leipein, to leave out, because the conic section falls short.

Greek also does something Latin does not. A Greek word beginning with a vowel could carry a rough breathing, a small h sound. When a prefix ending in a stop consonant ran into it, the stop turned into its aspirated partner: p became ph, t became th, k became ch. This is why several familiar words look mangled.

Greek prefixes that end in a vowel simply drop it before another vowel: ana becomes an- in aneurysm, anti becomes ant- in antagonist and antarctic, dia becomes di- in diocese, a- becomes an- before a vowel, which is why it is apathy and atypical but anaemia, anonymous and anarchy.

The two inventories, side by side

Latin and Greek supply English with two nearly complete and largely parallel sets of prefixes. Because English took both, it very often has two words for one idea, one from each side, differing only in register or in field.

Two things follow. A word mixing a Latin prefix with a Greek base is not an error; it is what a language with two toolkits does, and Lesson 16 takes up the argument. And when a technical term stalls you, its twin in the other language often will not: if diaphanous is opaque, transparent is not.

Common misconceptions

• Inflammable is the negative of flammable. It is not. The in- is directional, and flammable was promoted in the twentieth century precisely because the older word was being misread on the sides of fuel tankers.
• Every doubled consonant marks an assimilated prefix. Annual, irritate, terrible, collar and summit all carry an original double from the Latin base.
• in- always negates. Invaluable means beyond valuing, inhabitable means fit to live in, and ingenious has nothing to do with the absence of genius.
• Latin always wrote the assimilated form. Inscriptions and manuscripts show adfero beside affero and inpossibilis beside impossibilis for centuries. The assimilated spellings won slowly, and modern editors of Latin texts still differ over which to print.
• sur- is always super-. Surrogate and surreptitious are sub-. Only the sense of the whole word separates them.

The short version

• A prefix ending in a consonant tends to take on the consonant that follows it, which is why ad- appears as ac-, af-, ag-, al-, an-, ap-, ar-, as- and at-.
• in-, con-, sub-, ob-, ex-, dis- and inter- all do the same, and sub- and ob- sometimes drop their consonant instead of doubling it.
• English has two prefixes spelled in-, one negative and one directional, and only sense separates them; inflammable is the standing proof.
• A doubled consonant is a hypothesis, not a conclusion: annus, terrere, irritare and collum bring their own.
• Greek assimilates syn- and en- in the same way, and turns a stop into an aspirate before a rough breathing, giving method, cathode and ephemeral.
• The Latin and Greek prefix inventories run parallel, so most ideas have a word from each, and the familiar twin will often unlock the unfamiliar one.

Sources

1. Wikipedia. (n.d.). Assimilation (phonology). Wikimedia Foundation. en.wikipedia.org
2. Wiktionary. (n.d.). ad-. Wikimedia Foundation. en.wiktionary.org
3. Harper, D. (n.d.). Inflammable. Online Etymology Dictionary. etymonline.com
4. Allen, W. S. (1978). Vox Latina: A guide to the pronunciation of classical Latin (2nd ed.). Cambridge University Press.
5. Ayers, D. M. (1986). English words from Latin and Greek elements (2nd ed., revised by T. D. Worthen). University of Arizona Press.

Three words on one page of a contract

Open any commercial agreement and within a page you will meet an agent, an act and an agenda. They look like three unrelated pieces of business vocabulary. They are one Latin verb photographed at three moments: ago, agere, egi, actum, meaning to drive, to do, to conduct.

Agent is the present participle agens, the one doing. Agenda is the gerundive agenda, the things needing to be done. Act is the supine stem actum, the thing done. English took all three, at different dates and for different purposes, and now uses them without any sense that they are relatives.

Key idea: A Latin verb does not have one shape. It has several, and English borrowed from more than one of them. That is the reason the language holds both act and agent, both vision and video, both fact and efficient.

What a principal part is

Latin dictionaries list a verb under four forms, its principal parts, because from those four every other form can be generated. For ago they are:

• ago, first person singular present: I drive.
• agere, the infinitive: to drive.
• egi, first person singular perfect: I drove.
• actum, the supine, which is also the stem of the perfect passive participle actus: driven, done.

Only two of those matter for reading English. The present stem, here ag-, and the participial stem, here act-. The perfect egi almost never surfaces in English, which is a mercy, and the infinitive is just the present stem plus an ending.

Once you know that a verb has two stems, the apparently random shape of English word families snaps into order. Learn the pair and you get both halves of the family at once.

Ten verbs, two stems each

Two things in that table are worth stopping on. The present stem sometimes changes its vowel inside a compound: facio becomes -fic- in efficient and -fect- in perfect, capio becomes -cip- in recipient and -cept- in concept, teneo becomes -tin- in continent. That weakening is regular and Lesson 9 gives the rule. And the participial stem often swaps a t for an s after certain consonants, which is why mitto gives mission and video gives vision rather than mittion and vidtion.

The suffix knows which stem it wants

This is the part of the lesson you can use immediately. English suffixes are not free to attach anywhere. They arrived already fixed to one stem or the other in Latin, and they kept the habit.

So the pairs line up: agent and action, provident and provision, recipient and reception, component and composition, deficient and defect, consequent and execution. If you know one member you can build the other, and if a word looks wrong it usually is: there is no such English word as agention, because -ion refuses the present stem.

A small bonus falls out of this. Learners spend years guessing whether a word ends in -ant or -ent. The answer is in the Latin conjugation: verbs in -are give -ant (vacant, tolerant, important, dominant), and verbs in -ere and -ire give -ent (patient, evident, obedient, insistent). Attendant is from attendere, and so is an exception created inside French, but the rule holds far more often than chance.

The upshot: Two stems plus a short list of suffixes generates the whole family. This is why a reader who knows ten Latin verbs properly can read several hundred English words without having met them.

fero: one verb assembled out of two roots

The verb fero, ferre, tuli, latum, to carry or bear, looks broken, and it is. Its present is from one ancient root and its perfect and supine from others, exactly as English go supplies its own past tense from the unrelated went. Linguists call this suppletion.

The consequence for English is a family whose two halves have no letters in common. From fer-: refer, confer, prefer, transfer, infer, differ, offer, suffer, fertile, circumference, conifer, vociferous. From lat-: relate, translate, collate, prelate, superlative, dilatory, oblate, correlate, and, best of all, legislator, which is legis lator, a bearer of law, a person who proposes a statute rather than one who enforces it.

Two English words that feel unrelated are therefore near twins: to transfer and to translate both mean to carry across, one using the present stem and the other the supine.

Deponents, and the verb that hides inside execute

Some Latin verbs are passive in form and active in meaning. Grammarians call them deponent, from deponere, to lay aside, because they have laid aside their active forms. Their principal parts run to three, not four: sequor, sequi, secutus sum, meaning I follow.

English took both stems as usual. From sequ-: sequence, consequent, subsequent, sequel, obsequious, non sequitur. From secut-: execute, prosecute, persecute, consecutive. To execute is ex plus secutus, to follow a thing all the way out to its end, which explains why the word covers carrying out a plan, performing a piece of music, signing a deed and putting someone to death: all four are following something through to completion.

Other high-yield deponents work the same way. Loquor, loqui, locutus gives eloquent, loquacious and colloquial on one side, elocution, locution and interlocutor on the other. Patior, pati, passus gives patient on one side and passive, passion and compassion on the other, which is why the passion of a martyr and a patient in a ward are the same idea: someone undergoing.

Common misconceptions

• Video is a modern invention with no Latin behind it. It is the Latin present indicative, I see, taken into English in the 1930s on the model of audio, I hear. Vision, visible, visit and video are one family.
• The t in action belongs to the ending. It belongs to the supine stem actum. The suffix is -io, and the same -io appears in vision, where the stem ended in s.
• Every Latin verb yields exactly two English stems. Some yield one, and some yield three, because Latin also built repeated-action verbs on the supine: dicere gave dictare, which gave dictate and dictation alongside diction and dictum.
• An agenda is a list. Grammatically it is a plural, the things that must be done, and it entered English as a theological term for duties before it ever meant a sheet of paper.

What to carry forward

• A Latin verb is listed under four principal parts, and English draws almost entirely on two stems: the present stem and the participial or supine stem.
• Act and agent, vision and video, fact and efficient, captive and recipient are all one verb caught at two different moments.
• English suffixes are fixed to one stem or the other: -ent, -ence and -ile take the present stem; -ion, -or, -ure, -ive and -ory take the participial stem.
• The -ant or -ent question is usually settled by the Latin conjugation the verb belonged to.
• fero is suppletive, which is why refer and relate are the same verb with no letters in common, and why a legislator is a bearer of law.
• Deponent verbs such as sequor, loquor and patior behave normally for our purposes and give the usual two families, which is how execute and sequence turn out to be relatives.

Sources

1. Wikipedia. (n.d.). Principal parts. Wikimedia Foundation. en.wikipedia.org
2. Wiktionary. (n.d.). ago (Latin). Wikimedia Foundation. en.wiktionary.org
3. Lewis, C. T., & Short, C. (1879). A Latin dictionary. Clarendon Press. Digitised by the Perseus Digital Library. perseus.tufts.edu
4. Ayers, D. M. (1986). English words from Latin and Greek elements (2nd ed., revised by T. D. Worthen). University of Arizona Press.
5. Oxford English Dictionary. (2023). Entries for agenda, video and legislator (3rd ed., online). Oxford University Press. Subscription required.

Rex, and the missing x

Latin for king is rex. English derivatives include regal, regicide, regent, regime, regimen, regnant and interregnum. Not one of them has an x in it, and the vowel is different too. If you learned that rex means king and stopped there, none of those words would look like relatives.

The fix is one extra piece of information that every Latin dictionary prints and most vocabulary lists omit. The entry is not rex. It is rex, regis. The second word is the genitive, meaning of a king, and it shows the stem with the endings peeled back: reg-. Everything English built, it built on reg-.

Key idea: For a large class of Latin nouns the nominative singular is the one form that has been deformed by sound change. The genitive is undamaged, and the genitive is what English used.

Why the nominative lies

The nominative singular of the Latin third declension was formed by adding -s to the stem, and the collision that produced changed the stem. Take reg- plus -s: a g against an s comes out as the sound written x, so regs became rex. Take noct- plus -s: the t drops out and the same thing happens, giving nox. Take corpor- plus nothing, and the final r weakens to s in the nominative alone, giving corpus.

You do not need the phonology. You need the habit: whenever a Latin noun ends in -s, -x, -r or -us and the English derivatives look nothing like it, go and find the genitive.

Twenty nouns and the stems they hide

Read a few of those across and words stop being arbitrary. To record something is to put it back through the heart, which was where the Romans located memory. A mortgage is a dead pledge, because the obligation dies when the debt is paid. A manoeuvre is manu opera, work by hand, and so is the older sense of manure. To be excruciating is to be on the cross.

One entry in that list carries a trap worth having now. Latin has two nouns spelled os. One is os, oris, the mouth, which gives oral, orifice, oracle and peroration. The other is os, ossis, bone, which gives osseous, ossify and ossuary. The nominatives are identical and the genitives are not, which is a neat demonstration of why the genitive is the form that matters.

Adjectives, and the -x that turns into -c-

Adjectives behave the same way, and the pattern is easy to check because the English abstract noun in -ity always reveals the stem.

Nobody says feroxity. The -ity suffix goes on the stem, so the x of the nominative never appears, and the same is true of -ious, -ate and -ify. If you can produce the English abstract noun you already know the stem, which makes the process reversible: velocity tells you the stem is veloc-, and from there velocipede and velodrome are readable.

The last row hides a small marvel. Omnibus is not omni- plus bus. It is the whole Latin word, the dative plural of omnis, meaning for all. A Paris vehicle service advertised itself in the 1820s as a voiture omnibus, a carriage for everyone, and English clipped the word down to its Latin case ending. The bus you catch is the tail of a dative plural.

Worth holding on to: The English form in -ity, -ious or -ate is a window onto the Latin stem. When you cannot remember the genitive, build the abstract noun and read it off.

The -i- that shows up when Latin compounds

When Latin joins a noun or adjective stem to a second element, an -i- normally appears at the seam. That vowel is inherited, and it is the Latin counterpart of the Greek -o- you will meet in the next lesson.

• agr- plus -i- plus cultura gives agriculture; the same stem gives agrarian and, with per-, peregrinus, one who travels through the fields, which came into English through French as pilgrim.
• carn- plus -i- plus vorare gives carnivore; herb-, omn-, insect- and pisc- fill the same slot.
• magn- plus -i- plus facere gives magnify; the same shape produces pacify, fortify, clarify and nullify.
• frat(e)r- plus -i- plus caedere gives fratricide, and the -cide family runs through matricide, regicide, pesticide and suicide.
• mel, mellis, honey, plus -i- plus fluere gives mellifluous, flowing with honey.

Where the second element begins with a vowel the -i- usually drops out, which is why it is magnanimous rather than magnianimous, and quadrangle rather than quadriangle.

Two English words from one Latin stem

Several of these stems reached English twice: once as a learned borrowing straight from the Latin, and once by a slower road through French, wearing a different shape. The pairs are almost always split by register, the Latin one formal or technical and the French one ordinary.

Module 3 is about how that happened and what it did to the language. For now, notice only that the learned member of each pair sits closer to the stem, because it was copied from a page rather than transmitted through generations of speech.

Common misconceptions

• English builds on the form printed as the dictionary headword. It builds on the stem, which for third-declension nouns and adjectives only the genitive shows.
• os means bone. It means bone only if the genitive is ossis. If it is oris, the word means mouth, and oral and osseous have nothing to do with each other.
• Adjectives in -x keep the x in English. They never do: ferox gives ferocity, audax gives audacity, simplex gives simplicity.
• Omnibus is omni- plus a base bus. The whole word is a Latin dative plural meaning for all, and bus is what English kept when it threw away the front of it.
• The connecting -i- is part of the base. It is a joint. Strip it and agr-, carn-, magn- and fratr- stand up on their own.

Where this leaves us

• The nominative singular of a third-declension Latin noun is the form most deformed by sound change; the genitive shows the stem English actually borrowed.
• rex regis gives reg-, lex legis gives leg-, corpus corporis gives corpor-, iter itineris gives itiner-, and each stem then generates a family.
• Two different nouns are spelled os, and only the genitive separates the mouth from the bone.
• Adjectives follow the same rule, and the English abstract noun in -ity is a reliable window onto the stem.
• Latin joins elements with -i-, which drops before a vowel; that vowel is the Latin counterpart of the Greek -o-.
• Many stems arrived twice, once learned and once through French, leaving pairs such as regal and royal that differ chiefly in register.

Sources

1. Wikipedia. (n.d.). Latin declension. Wikimedia Foundation. en.wikipedia.org
2. Wiktionary. (n.d.). corpus (Latin). Wikimedia Foundation. en.wiktionary.org
3. Lewis, C. T., & Short, C. (1879). A Latin dictionary. Clarendon Press. Digitised by the Perseus Digital Library. perseus.tufts.edu
4. Ayers, D. M. (1986). English words from Latin and Greek elements (2nd ed., revised by T. D. Worthen). University of Arizona Press.
5. Green, T. M. (2014). The Greek and Latin roots of English (5th ed.). Rowman and Littlefield.

Three words and a vowel nobody put there

A thermometer measures heat. A thermostat holds it steady. A Thermos, a trade name registered in 1904, keeps it. The Greek word behind all three is therme, heat, whose stem is therm-. The word for measure is metron, whose stem is metr-. Neither of them contains the o that sits between them in thermometer.

That o is a joint, and it is the single most useful diagnostic in the language. Latin joins its elements with -i-: agriculture, carnivore, magnify. Greek joins with -o-: thermometer, psychology, geography, chronology. When a technical word has an -o- at the seam, you are almost certainly looking at Greek.

Key idea: Greek did not lend English words so much as parts. Those parts are called combining forms, and they are still being fitted together to make words that no Greek ever spoke.

The forms worth knowing

Thirty forms and a handful of prefixes will carry you through a startling amount of technical English. Notice that several stems end in a consonant that only shows up in the derivatives: phos, light, has the stem phot-, which is why it is photograph and not phosograph, and onoma, name, has the stem onomat-, which is why onomatopoeia has a t in it.

When the joining vowel stays and when it goes

The -o- is not automatic. Medical terminology, which builds more new words from Greek than any other field, states the rule in three parts, and it holds well outside medicine.

1. Keep the -o- before a suffix beginning with a consonant. cardi-o-logy, gastr-o-scopy, oste-o-porosis, hepat-o-megaly.
2. Drop it before a suffix beginning with a vowel. gastr-itis, not gastroitis; arthr-itis; hepat-itis; neur-algia; append-ectomy; cardi-ac.
3. Keep it between two bases, even when the second begins with a vowel. gastr-o-enteritis, oste-o-arthritis, gastr-o-oesophageal.

Rule three is why gastroenteritis keeps its o while gastritis loses it, a contrast that looks arbitrary until you notice that enter- is a base and -itis is a suffix.

The point: Latin joins with -i-, Greek with -o-, and the Greek vowel survives before a consonant, vanishes before a vowel suffix, and holds firm between two bases.

Greek suffixes

A short list covers most of what you will meet. The condition suffix -ia gives amnesia, euphoria and dyslexia. The process suffix -sis gives analysis, synthesis, diagnosis and metamorphosis, with adjectives in -tic. The result suffix -ma has a stem in -mat-, which is why dogma becomes dogmatic, drama dramatic, schema schematic and trauma traumatic. The resemblance suffix -oid, from eidos, form, gives asteroid, rhomboid and humanoid. And the trio -ist, -ism and -ize, from Greek verbs in -izein, is now so productive that English attaches it to anything at all.

How Greek was written down in Latin, and then in English

English did not take Greek words from Greek. It took them from Latin, and Roman writers of the first century BCE had settled on a fixed way of transliterating them. Those conventions are the reason Greek words look the way they do in English, and they let you spot one across a page.

Put the row for upsilon together with the rows for ph, th, ch and rh and you have a working detector. A learned English word containing y in the middle, or ph, or ch pronounced as k, or an initial ps-, pn-, mn-, pt- or x-, is Greek far more often than not. Pneumonia, mnemonic, pterodactyl, ctenophore, xylophone and xenon all keep a cluster that English cannot pronounce and so silences.

There is one honourable exception worth knowing so it does not trip you: nephew has a ph and is not Greek. Middle English spelled it nevew, from French neveu and Latin nepos; sixteenth-century scholars respelled it on a mistaken hunch. That kind of learned meddling is the subject of Lesson 10.

Why oeconomics became economics, and why foetus was never Greek

Greek oikonomia is oikos, household, plus nomos, management: the running of a household. Latin wrote it oeconomia, and English wrote oeconomy for two centuries. Samuel Johnson's Dictionary of 1755 has it under OECONOMY. The digraph was simplified across the nineteenth century, faster in the United States under Noah Webster's influence and eventually in Britain too, leaving only a small British-held remnant: amoeba, oesophagus, oestrogen, diarrhoea, and the French intruder manoeuvre, which is not Greek at all.

The same tidying explains why a British paediatrician and an American pediatrician do the same job, and why aesthetic and esthetic are one word. What it does not explain is foetus, and this is the case worth remembering. Foetus has no Greek in it. The Latin word is fetus, from a root meaning bearing or offspring, the same root that gives fecund, fertile and female. The spelling with oe was introduced by early modern writers who assumed a Greek digraph that had never been there. Modern anatomical and editorial style has largely gone back to fetus, and the Oxford English Dictionary records the oe form as an error of learning.

One more spelling myth belongs here. The ending -ize is not an Americanism. It renders the Greek verb ending -izein through Latin -izare, and it is the spelling the Oxford English Dictionary and Oxford University Press have used for well over a century. The -ise spelling reached English through French. Both are perfectly good; only the history is commonly told backwards.

Common misconceptions

• The -o- in photo- and hydro- belongs to the root. It is a joint. The roots are phot- and hydr-, and the vowel disappears the moment a vowel suffix follows.
• ae and oe are British spellings. They are Latin renderings of Greek digraphs. Britain kept them and American spelling simplified them, and both are still simplifying.
• foetus is the correct traditional spelling. Classical Latin has fetus. The oe is an early modern hypercorrection by writers who thought the word was Greek.
• -ize is an American import. It is older than -ise and closer to the Greek. Oxford has used it throughout.
• Any y in the middle of a word marks a Greek root. Only in the learned layer. Very, carry and marry have y for entirely different reasons.

Pulling it together

• Greek lent English combining forms rather than finished words, and about thirty of them carry most of the technical vocabulary.
• The joining vowel is -o- in Greek and -i- in Latin, which is the fastest single test of which language a compound comes from.
• Keep the -o- before a consonant suffix, drop it before a vowel suffix, and keep it between two bases: cardiology, gastritis, gastroenteritis.
• Some stems differ from the dictionary form: phos has the stem phot-, onoma has onomat-, and the derivatives show it.
• Roman transliteration fixed kappa as c, chi as ch, phi as ph, theta as th, upsilon as y and the rough breathing as h, and English inherited all of it.
• oeconomy became economy by nineteenth-century simplification; foetus never had an oe to simplify, and the spelling is a scholarly mistake.

Sources

1. Wikipedia. (n.d.). Classical compound. Wikimedia Foundation. en.wikipedia.org
2. Wikipedia. (n.d.). Romanization of Greek. Wikimedia Foundation. en.wikipedia.org
3. Harper, D. (n.d.). Economy. Online Etymology Dictionary. etymonline.com
4. Liddell, H. G., Scott, R., & Jones, H. S. (1940). A Greek-English lexicon (9th ed.). Clarendon Press. Digitised by the Perseus Digital Library. perseus.tufts.edu
5. Oxford English Dictionary. (2023). Entries for fetus and -ize (3rd ed., online). Oxford University Press. Subscription required.

18 Germinal, Year III

On 7 April 1795, a date the French Republic called 18 Germinal of Year III, the National Convention passed the law that gave the metric system its vocabulary. The units were new and so were their names, and the drafting committee made a decision that the world has lived with ever since: multiples of the base unit would take Greek prefixes, and fractions of it would take Latin ones.

So the law lists myria- for ten thousand, kilo- for a thousand, hecto- for a hundred and deca- for ten, all Greek, against deci- for a tenth, centi- for a hundredth and milli- for a thousandth, all Latin. A reader who knows the two number systems can therefore tell at a glance which way a metric prefix points, without knowing anything about the unit it is attached to. That was the intention, and for about a century and a half it held.

Key idea: English counts in two dead languages at once. The Latin set and the Greek set overlap in meaning and almost never in form, and knowing which is which resolves a surprising number of otherwise arbitrary spellings.

The two sets, side by side

Three is the same in both because both languages inherited it from a common ancestor, which is why tricycle and tripod feel like the same word and are. A mile is mille passuum, a thousand paces, each pace being two steps of a marching legionary. A hecatomb was a sacrifice of a hundred oxen, and the word has kept its scale even though nobody has performed one in a very long time.

The table also explains a small family quarrel. A nine-sided figure ought to be an enneagon if the whole word is Greek, since -gon is Greek gonia, angle. Almost nobody says enneagon. Everybody says nonagon, which welds a Latin nine onto a Greek angle. The purists lost that one too, and Lesson 16 takes up why.

September is the seventh month

Line the month names up against the numbers and they are wrong by two. September is sept-, seven. October is oct-, eight. November is nov-, nine. December is dec-, ten. All four sit two places later in the calendar than their names claim.

The explanation is that the early Roman calendar began the year in March, so the numbered months counted from there. January and February were added at the front, and the numbered names were never adjusted. Julius Caesar's reform of 46 BCE and the later renaming of two months after Julius and Augustus left the arithmetic exactly as broken as it had been.

Remember: A word can go on carrying a fact that has stopped being true. That is a warning about etymology as much as about calendars, and Lesson 10 returns to it.

Halves, and one and a half

English has three prefixes for half and uses all of them. Semi- is Latin, hemi- is Greek, and demi- is French, itself worn down from Latin dimidius. They are not interchangeable in practice: semicircle, hemisphere and demigod each keep the language of the base they are attached to, mostly.

Musical notation manages to use all three in a single word. A quaver is an eighth note. Halve it and you have a semiquaver, a sixteenth. Halve that and you have a demisemiquaver, a thirty-second. Halve it once more and you have a hemidemisemiquaver, a sixty-fourth note built from Greek, French, Latin and an English word for trembling.

Latin also had a prefix for one and a half, sesqui-, contracted from semis-que, and a half in addition. A sesquicentennial is a hundred and fiftieth anniversary. Sesquipedalian, foot-and-a-half-long, is Horace's joke from the Ars Poetica, where he mocks poets who throw around sesquipedalia verba, words a foot and a half long. It is now itself a sesquipedalian word, which is the point.

The metric prefixes, and how the logic broke

The 1795 scheme was tidy. What followed was not, because as science needed larger and smaller numbers the committees ran out of classical vocabulary and improvised.

Read the right-hand column downwards and you can watch a rule dissolve. It begins as Greek for large and Latin for small. Then Greek is used in both directions. Then an Italian word is drafted in. Then two Danish numerals, chosen because fifteen and eighteen were the exponents needed. Then Latin and Greek numerals for the powers of a thousand, disguised with a spare consonant so that no two prefixes share a symbol. Then, in November 2022, four names invented to use the letters that were left.

The practical consequence is that you cannot read a modern metric prefix off its language. You can read the four oldest pairs, and after that you are memorising.

Kilobytes, and the second muddle

Computing borrowed kilo-, mega- and giga- and quietly changed their values, because powers of two are convenient in memory and 2 to the tenth is 1024, close enough to a thousand to reuse the word. For decades a kilobyte was 1024 bytes to a programmer and 1000 bytes to a disk manufacturer, which is why a drive sold as 500 gigabytes shows up as roughly 465 in some operating systems.

In 1998 the International Electrotechnical Commission proposed a separate set for the binary values: kibibyte, mebibyte, gibibyte and tebibyte, built by replacing the second syllable of the Greek prefix with bi, for binary. They are correct, unambiguous and largely ignored outside standards documents, which is a fair summary of what happens when a committee tries to fix a word that people are already using.

Common misconceptions

• September means the ninth month. It means the seventh. The Roman year began in March and the numbered names were never corrected.
• Decimate can only mean to reduce by one tenth. The Roman military punishment did kill one man in ten, but English has used the word for heavy destruction since the seventeenth century. Insisting on the older sense is the etymological fallacy, which Lesson 10 takes apart.
• The plural of octopus is octopi. Octopus is Greek, so a Latin plural in -i is wrong twice. Octopuses is standard English and octopodes is the pedant's form.
• All the metric prefixes are Greek. Only the 1795 multiples were. The submultiples were Latin, pico- is Italian, and femto- and atto- are Danish.
• A kilobyte is 1024 bytes. In the metric system it is 1000. The binary value has its own name, kibibyte, which almost nobody uses.

Summing up

• English carries two full sets of numeral elements, Latin and Greek, that overlap in meaning and diverge in form except at three.
• The law of 7 April 1795 assigned Greek prefixes to metric multiples and Latin ones to submultiples, and that division still explains kilo- against milli-.
• The scheme broke as science needed more range: pico- is Italian, femto- and atto- are Danish, and the 2022 additions were built around spare letters.
• September through December are two months out because the Roman year once began in March.
• Semi-, hemi- and demi- are Latin, Greek and French for half, and the hemidemisemiquaver uses all three at once.
• sesqui- means one and a half, and sesquipedalian is Horace's joke about words a foot and a half long.
• Computing reused kilo-, mega- and giga- for powers of two, and the corrective names kibi- and mebi- never caught on.

Sources

1. Bureau International des Poids et Mesures. (n.d.). SI prefixes. BIPM. bipm.org
2. National Institute of Standards and Technology. (n.d.). Metric (SI) prefixes. NIST Office of Weights and Measures. nist.gov
3. Wikipedia. (n.d.). Metric prefix. Wikimedia Foundation. en.wikipedia.org
4. Harper, D. (n.d.). Decimate. Online Etymology Dictionary. etymonline.com
5. Horace. (c. 19 BCE). Ars poetica, line 97 (sesquipedalia verba). In Satires, epistles and Ars poetica (H. R. Fairclough, Trans., 1926). Loeb Classical Library, Harvard University Press.

Two words for easily broken

English has frail and it has fragile. They mean nearly the same thing, they behave differently in a sentence, and they are the same Latin word. Fragilis went into spoken Latin, wore down through Gaulish speech into Old French fraile, and reached English in the thirteenth century as frail. Then, around 1600, English scholars reading Latin took fragilis again, straight off the page, and wrote fragile.

Three hundred years apart, one etymon, two English words. Pairs like these are called doublets, and English has more of them than almost any language in Europe, because it kept letting Latin in by different doors.

Key idea: The same Latin word can be in English several times over, in several shapes, because it arrived by several routes at several dates. Deciding which route a word took usually tells you what it now sounds like: plain, middling or learned.

Door one: Latin into Old English

The first Latin words in English arrived before English was in England. Germanic-speaking peoples traded with the Roman empire, and the vocabulary that crossed is exactly what you would expect from trade and settlement: wine from vinum, cheese from caseus, street from strata via, a paved way, wall from vallum, mile from mille, pound from pondo, cup from cuppa, kitchen from coquina, pepper from piper, mint from moneta, the temple of Juno Moneta where Rome struck its coins.

A second wave came with the conversion from 597 onward, and it is religious and scholarly: bishop from episcopus, priest from presbyter, monk from monachus, altar, mass from missa, angel and devil, both Greek by way of Latin, and school from schola. Church is thought to come from Greek kyriakon, the Lord's house, through Germanic rather than through Latin.

Every one of those words now looks and sounds native. They have been in the language for over a thousand years, they take native inflections, and no English speaker feels them as foreign. That is what a very old borrowing looks like.

Door two: through French

After 1066 the ruling class of England spoke a variety of Norman French, and for three centuries the language of government, law, the court and high cuisine was French. English took tens of thousands of words from it.

The date is where most accounts go wrong. Borrowing did not spike in 1066. The Norman elite spoke French and had no reason to translate; large-scale borrowing happened as French speakers shifted to English and carried their vocabulary across with them. Philip Durkin's count of first attestations in the Oxford English Dictionary puts the peak in the fourteenth century, more than two hundred years after the Conquest.

Because French had already reshaped its Latin inheritance, these words look worn: royal from regalis, loyal from legalis, chief from caput, count from computare, sure from securum. And because English was taking from two French dialects at different times, it collected pairs that differ by dialect rather than by date.

Warranty and guarantee are therefore not two ideas but one word entering twice from two dialects, which is why lawyers still use both and cannot quite say what separates them.

Door three: taken again, off the page

From roughly 1500 to 1700, English writers reached into Latin and Greek directly for anything the language seemed to lack. Some of it stuck: fragile, regal, hospital, factory, species, radius, genus, formula, apparatus, appendix, focus, series, index, medium. Some of it did not: the period produced deruncinate, for to weed out, and illecebrous, for alluring, and a great many other words that got one outing apiece.

Contemporaries argued about it. Critics called the new words inkhorn terms, after the ink pot, meaning bookish importations for showing off. The argument was not settled so much as outlived: the useful borrowings survived and the show-offs did not, and nobody now hears fragile or appendix as pretentious.

What matters here: Door three words sit closest to the Latin stem, because they were copied rather than transmitted. That is why regal is nearer to regalis than royal is.

The doublets themselves

The capital row deserves a sentence of its own. Latin capitale meant property, from caput, head, because wealth was counted in head of livestock. That single word gives modern English cattle, the animals themselves; chattel, a lawyer's word for movable property; and capital, the money. Three English words, one Latin idea, and the accounting sense visible in all three.

Chief and chef are the same word too. Chief came in from Norman French in the thirteenth century meaning head or leader; chef came in from modern French in the nineteenth as a clipping of chef de cuisine, the head of the kitchen. The gap between them is six hundred years of French sound change.

Three doors, three registers

Because the doors opened at different social levels, English ended up with sets of near-synonyms that differ mainly in how formal they sound. The native or door-one word is plain, the French word is middling, and the learned Latin or Greek word is technical.

This is why English prose can change temperature without changing subject, and why a writer choosing between end and conclude is choosing a social register rather than a meaning. It is also a practical tool: if a learned word defeats you, the native word in its row usually will not.

Common misconceptions

• French vocabulary flooded English in 1066. The Conquest set up the conditions; the borrowing peaked in the fourteenth century, as French speakers shifted to English.
• Latin words are the formal ones. Wine, cheese, street, wall, mile, cup, kitchen and pound are all Latin, and all of them are as plain as English gets. Age of borrowing matters more than source.
• In a doublet, one member is the real word and the other is a variant. Both descend from the etymon. Frail is not a corruption of fragile; they are siblings who took different roads.
• Warranty and guarantee differ in meaning by design. They are one word from two French dialects, and the legal distinction between them was invented afterwards to justify keeping both.
• The inkhorn controversy was won by one side. It was decided word by word by use. Fragile survived; illecebrous did not.

What you now know

• Latin entered English by three main routes: directly into Old English, through French after 1066, and by learned reborrowing from about 1500.
• A doublet is one etymon arriving twice; frail and fragile are three hundred years apart, and hostel, hotel and hospital are a triplet.
• Norman and Central French supplied pairs that differ by dialect, giving warranty beside guarantee and cattle beside chattel.
• capitale, property counted in head of livestock, gave English cattle, chattel and capital at once.
• The learned member of a pair sits closest to the Latin stem because it was copied from a page rather than passed through speech.
• The three doors produced the three-tier register of English synonyms, from ask through question to interrogate.

Sources

1. Wikipedia. (n.d.). Doublet (linguistics). Wikimedia Foundation. en.wikipedia.org
2. Wikipedia. (n.d.). Latin influence in English. Wikimedia Foundation. en.wikipedia.org
3. Harper, D. (n.d.). Cattle. Online Etymology Dictionary. etymonline.com
4. Durkin, P. (2014). Borrowed words: A history of loanwords in English. Oxford University Press.
5. Baugh, A. C., & Cable, T. (2013). A history of the English language (6th ed.). Routledge.

A plausible answer that falls apart

Father, and Latin pater. Both mean father. Both begin with a lip consonant, both have a t in the middle, both end in a syllable spelled with r. The obvious conclusion is that English took the word from Latin, the way it took paternal, patriarch and patron. That conclusion is wrong, and finding out exactly where it fails is one of the more useful things a word-builder can do.

Start with the date. Father is in the oldest English we have, as faeder, in texts from before the Latin learned borrowings began. It has relatives in every Germanic language: German Vater, Dutch vader, Old Norse fadir, Gothic fadar. A word borrowed from Latin into English would not be sitting in Gothic, a language whose speakers had no such contact and which died out around the ninth century.

Then look at the consonant. Every one of those Germanic forms has f or v where Latin has p. If English had borrowed the word, English would have a p in it, exactly as it does in paternal. A borrowing copies the sounds it hears. An inheritance changes them, according to rules, over centuries.

Key idea: Father and paternal are relatives, but the relation is cousinhood, not descent. Both go back to a shared ancestor, and the Germanic branch changed its consonants on the way.

The consonants that give it away

The changes are described by Grimm's law, published by Jacob Grimm in 1822 on the basis of Rasmus Rask's earlier work. LING 340 in this catalogue works the law out properly, with the counter-examples that Verner's law resolved in 1875. Here we want only the part that helps you read a word: a table of what Latin and Greek kept and what Germanic changed.

Read that table once and a great many word pairs stop being coincidences. Greek behaves like Latin for our purposes: pous, podos beside foot, treis beside three, deka beside ten, hekaton beside hundred, kardia beside heart, odous, odontos beside tooth.

The test, as a procedure

Here is the working version. Take an English word and a classical word that resemble each other, then run three steps.

1. Compare the consonants. If the English word has f where the classical word has p, th where it has t, h where it has c or k, t where it has d, or k where it has g, the two are cousins, inherited separately.
2. If the consonants match exactly, English almost certainly borrowed the word, whether directly or through French.
3. Check the age. An inherited word will be short, common, and present in the other Germanic languages. A loan will be longer, more specialised, and datable.

Run it on a few triples and you can see the whole architecture of the English vocabulary at once.

The upshot: Each row is one ancient word that entered English three times: once by inheritance, once by loan from Latin, once by loan from Greek. The plain word, the formal word and the technical word are all the same word, wearing three different amounts of history.

Two pairs that look related and are not

The test cuts the other way as well, and its best results are the ones that refuse an obvious answer.

Day and dies. They mean the same thing and look alike, and they are unrelated. Latin dies goes back to a root meaning to shine, the one behind Iuppiter and divine. English day goes back to a different root, one meaning to burn or be warm. The resemblance is an accident, and the consonants do not line up in the way the table requires.

Have and habere. Here the accident is even better. English have does not come from Latin habere. Apply the table: English h corresponds to classical k, so have is the cousin of Latin capere, to take, and therefore the relative of capture, captive and receive. Meanwhile Latin habere goes back to a sound that gives Germanic g, so its actual English cousin is give. English have matches Latin take, and English give matches Latin have. Nothing about the spelling tells you this. The correspondences do.

The layer English never borrowed

Alongside the classical machinery, English kept its own set of word-building parts, and they are not interchangeable with the borrowed set.

• Native suffixes: -ness (kindness), -hood (childhood), -ship (friendship), -dom (freedom), -ful, -less, -ly, -ish, -er, -en, and the -th of warmth, length, depth and truth.
• Native prefixes: un-, over-, under-, out-, fore-, mis-, be-, with-, off-.
• Free compounding: bookshop, doorstep, handbook, sunrise. No connecting vowel, no bound bases, no rules to learn.

Put the two systems side by side on the same idea and you get the register difference that English writers live on.

Notice that both members of each pair are built from the same materials, a base plus an adjective suffix. The difference is only which stockroom the parts came from, and the reader hears it instantly.

There is a further point about proportions that is easy to state backwards. Count entries in a large dictionary and English looks overwhelmingly Latin and Greek. Count the words in an ordinary paragraph and it looks overwhelmingly Germanic, because the pronouns, prepositions, auxiliaries and the hundred commonest words are all native. Both counts are correct. The classical layer supplies most of the vocabulary and almost none of the running text.

Common misconceptions

• Father comes from Latin pater. It is inherited, not borrowed, and the f against the p is the proof. Paternal is the borrowing.
• Have comes from habere. Have is the cousin of capere, and give is the cousin of habere. The correspondences cross over.
• Day comes from Latin dies. Two different ancient roots that happen to have converged in shape.
• Grimm's law has too many exceptions to trust. Verner accounted for the largest class of them in 1875 by showing that the position of the ancient accent governed the outcome, which is exactly what turned the law from a pattern into a law.
• The Latinate word is the better word. It is the more formal word. Motherly and maternal describe different situations, not different degrees of correctness.

Looking back

• Father is inherited and paternal is borrowed; both descend from one ancient word by different histories.
• Germanic changed p to f, t to th, k to h, d to t and g to k, so a mismatch of exactly that kind marks a cousin rather than a loan.
• The same ancient word often reached English three times, giving triples such as heart, cordial and cardiac.
• Day and dies are unrelated, and have is the cousin of capere while give is the cousin of habere.
• English kept a full native word-building system, with -ness, -hood, -dom, un- and free compounding, which never adopted connecting vowels or bound bases.
• Native and classical words for the same idea differ in register, and English prose uses that difference constantly.

Sources

1. Wikipedia. (n.d.). Grimm's law. Wikimedia Foundation. en.wikipedia.org
2. Wikipedia. (n.d.). Verner's law. Wikimedia Foundation. en.wikipedia.org
3. Harper, D. (n.d.). Father. Online Etymology Dictionary. etymonline.com
4. Sihler, A. L. (1995). New comparative grammar of Greek and Latin. Oxford University Press.
5. Ringe, D. (2017). From Proto-Indo-European to Proto-Germanic (2nd ed.). Oxford University Press.

Seven words, one verb

Confection, efficient, difficult, perfect, artifice, sacrifice, benefactor. Seven English words, and one Latin verb underneath all of them: facio, facere, feci, factum, to make or do. The a of facio survives in benefactor and artifice. It has become an i in efficient and difficult, and an e in perfect and confection.

None of that is random. Latin changed its own vowels according to a rule, and if you know the rule you can run it backwards and find the verb. This lesson is three Latin changes and four French ones, and then a procedure that undoes all seven.

Key idea: A base you cannot see is usually a base that has been altered by a regular process. Regular processes can be reversed.

Change one: the vowel in the middle weakens

Very early Latin stressed the first syllable of every word hard, and the vowels in the syllables after it wore down. When a prefix was added, the base's first syllable stopped being first, and its vowel weakened accordingly. The outcome is regular enough to tabulate.

• a becomes i in an open syllable: facio gives conficio, capio gives recipio, ago gives exigo, habeo gives prohibeo, statuo gives constituo.
• a becomes e in a closed syllable: factus gives confectus, captus gives acceptus, raptus gives correptus.
• e becomes i: teneo gives contineo, lego gives colligo, premo gives comprimo, sedeo gives obsideo.
• ae becomes i: quaero gives inquiro, caedo gives incido.
• au becomes u: claudo gives includo, causa gives accuso.

Two rows of that table look identical and are not. Cado, to fall, and caedo, to cut, both come out as -cido in compounds, which is why incident and incision seem to belong together. They do not. An incident is something falling upon you; an incision is a cut into something. The whole -cide family, homicide, pesticide, regicide, comes from caedo, and the whole -cidence family, coincidence, incidence, comes from cado. The participles keep them apart: -casus against -cisus.

The point: When a base looks unfamiliar, try restoring the vowel. An i in a syllable after a prefix was very often an a.

Change two: the participle's t turns into s

When the participial ending -tus met a base ending in d, t or certain other consonants, the cluster simplified to s or ss. That is why English has vision rather than a form in -t-, and mission rather than mittion.

Change three: s between vowels turns into r

Some time in the fourth century BCE, Latin turned s between vowels into r. The change is called rhotacism, and Cicero preserves a note about it: he tells his correspondent Paetus that a man of the Papirius family was originally called Papisius, and that the name changed with the sound.

The effect on English is everywhere. Latin nouns whose nominative ends in -s show r throughout the rest of the paradigm, because in the nominative the s was final and safe, while in every other case it stood between vowels and became r.

• flos, floris: flower survived as flor-, giving floral, florid, flora and efflorescence.
• honos, honoris: honorary, honorific. The older honos is still printed in some editions of Cicero.
• genus, generis; corpus, corporis; opus, operis; tempus, temporis: general, corporeal, operate, temporal.
• ius, iuris: jury, jurisdiction, jurisprudence. But justice keeps its s, because there the s was never between vowels.
• mus, muris: murine, a mouse.

English did the same thing independently in its own prehistory, which is why the language says I was but we were, and he is but they are. Lesson 8's cousins-not-loans point applies here too: two languages made the same change without either copying the other.

What French wore away

The Latin that reached English through French had already been through several centuries of erosion. Four changes account for most of the disguise.

The third row is the whole explanation of the hostel and hotel doublet. English borrowed the French word while the s was still there and got hostel; it borrowed the same word again after the s had gone and got hotel. The circumflex on the modern French spelling is not a length mark. It is a headstone for a lost s, which is why hopital, foret and ile line up with hospital, forest and insular.

The procedure, run backwards

Now put all seven changes together into something you can use on a word you have never seen.

1. Strip the English suffix and any inflection.
2. Un-assimilate the prefix, using Lesson 2.
3. Restore the weakened vowel: an i or e in the syllable after a prefix was often an a; an i may have been ae; a u may have been au.
4. If the ending is -sion or -ss-, try putting a t or d back.
5. If an r sits between vowels in a Latin stem, try an s.
6. If the word looks French-worn, put -ct- back for -it-, ca- for cha-, an s back before a consonant, and a consonant back between the vowels.
7. Check the result in a dictionary. The procedure gives you a hypothesis, not a fact.

Three worked examples. Surreptitious. Step two: sur- is sub-. Step three: -rept- restores to rapt-, from rapio. So the word is snatched from underneath, which is exactly what it means. Exquisite. Step three: -quisit- restores to quaesit-, from quaero. Sought out, hence choice. Trait. Step six: -ait restores to -act-, giving tract-, from trahere, to drag. A trait is a drawn line, which is also what a portrait is: a line drawn forth.

Common misconceptions

• The -cide of homicide and the -cidence of coincidence are the same element. They are two verbs, caedo to cut and cado to fall, that happen to weaken to the same shape.
• Rhotacism means mispronouncing r. In speech therapy it does. In historical linguistics it means an s becoming an r, which is what happened inside Latin.
• French corrupted Latin. French continued Latin. The forms are the regular outcomes of ordinary sound change, no more corrupt than father is a corruption of pater.
• The circumflex in hotel and foret marks a long vowel. It marks a deleted s, which the English cognates hostel and forest still carry.
• Vision must contain a suffix -sion. The s belongs to the participial stem visus; the suffix is -io.

The takeaway

• Latin weakened medial vowels after a prefix, so facio becomes -ficio and factus becomes -fectus; running that backwards recovers the base.
• cado and caedo both give -cido, which is why incident and incision look related and are not.
• The participial -tus became -sus after certain consonants, giving vision, mission, session and pressure.
• Rhotacism turned s between vowels into r, which is why flos has the stem flor- and ius has iuris while justice keeps its s.
• French turned -ct- into -it-, ca- into cha-, dropped s before consonants and dropped consonants between vowels.
• The circumflex of hotel is a lost s, and hostel is the same word borrowed before it went.
• The seven changes reverse into a procedure that turns an opaque word into a checkable hypothesis.

Sources

1. Wikipedia. (n.d.). Rhotacism (sound change). Wikimedia Foundation. en.wikipedia.org
2. Wiktionary. (n.d.). conficio (Latin). Wikimedia Foundation. en.wiktionary.org
3. Wikipedia. (n.d.). Old French. Wikimedia Foundation. en.wikipedia.org
4. Cicero. (c. 46 BCE). Epistulae ad familiares 9.21 (on Papisius and Papirius). In Letters to friends (D. R. Shackleton Bailey, Ed. and Trans., 2001). Loeb Classical Library, Harvard University Press.
5. Sihler, A. L. (1995). New comparative grammar of Greek and Latin. Oxford University Press.

A crayfish is not a fish

Old French had a word escrevisse, later crevice, for the small freshwater lobster. It is Germanic in origin and a relative of German Krebs, a crab. English took it in the fourteenth century, heard the second syllable as a familiar word, and wrote crayfish. In parts of the United States the same reanalysis went a different way and produced crawfish, which at least has the animal's gait in it.

The creature did not change. The word did, because speakers made sense of an unfamiliar shape by seeing a familiar one inside it. That process is called folk etymology, and it is the single most reliable way for a confident parser to be confidently wrong.

Key idea: A word that comes apart neatly is not thereby explained. Neatness is what folk etymology manufactures.

Five words that were rebuilt by their readers

• Belfry has no bell in it. Old French berfrei, a movable siege tower, from a Germanic compound meaning protect-and-shelter. English changed the first r to l, and because such towers held bells, nobody has been able to unhear the connection since.
• Island is native English, Old English igland, and had no s for six hundred years. In the sixteenth century, scholars who knew Old French isle and Latin insula added one to a word that was not related to either. The learned s is a mistake by educated people, which is worth remembering when folk etymology is described as a peasant's error.
• Female is not built on male. It is Old French femelle, from Latin femella, a small woman, and it reached English as femelle. Male is a different word entirely, from Latin masculus by way of Old French masle. The two converged in spelling because English speakers assumed a relationship the Latin never had.
• Hangnail does not hang. Old English angnaegl was a painful spike or corn, with ang- meaning tight or painful, the same element as in anguish and angina, which Lesson 1 met on a drug label.
• Bridegroom has no groom. Old English brydguma is bride plus guma, a man, which is the native cousin of Latin homo. When guma dropped out of the language the word made no sense, so the nearest available word was fitted into the gap.

The list runs on. Sirloin is Old French surloigne, above the loin, and no king ever knighted a joint of beef, though the story has been told about three different kings. A forlorn hope is Dutch verloren hoop, a lost troop, and the hoop is a heap of soldiers, not an expectation. Penthouse is Old French apentis, an appended lean-to, from the same Latin as appendix. Cockroach is Spanish cucaracha. Helpmeet was created by misreading the King James phrase an help meet for him, where meet is an adjective meaning suitable.

Elements that look identical and are not

The second trap is inside the classical vocabulary rather than outside it. Some elements are two elements wearing one spelling.

The two -logies. Greek legein means both to speak and to gather, and both senses fed the noun logos and the ending -logia. So biology, geology and astrology use the speaking sense, the account or study of a thing. But a trilogy is a set of three, a tetralogy a set of four, and an anthology is anthos plus -logia, a gathering of flowers. Nobody studies flowers in an anthology, and no one studies threes in a trilogy.

Astrology carries a second lesson. It and astronomy were interchangeable words for centuries, and astrologia was often the more respectable of the two. The modern division of labour, with astronomy taking the motions and astrology the influences, hardened only in the seventeenth century as the second discipline lost its standing.

The four -ates. The ending does four different jobs. It makes verbs from Latin participles, as in separate and activate. It makes adjectives, as in fortunate and temperate. It names an office or a body of people, as in doctorate, electorate, magistrate and senate. And in chemistry it names a salt of an acid, as in sulphate, nitrate and carbonate. A reader who assumes one -ate will misread three.

The two in-. Lesson 2 covered this, and it belongs on any list of false friends: inflammable and invaluable use the directional prefix, not the negative one.

-ic against -ical. Often interchangeable, and occasionally not. An economic decision is about the economy; an economical one saves money. A historic day makes history; a historical day merely happened in the past. Nothing in the parts tells you which; only use does.

Words that keep a dead theory alive

A third class of trap is a word whose etymology is completely correct and completely misleading, because the idea it encodes has been abandoned.

These are worth keeping because they are honest fossils. They tell you what people thought at the moment the word was coined, and nothing whatever about what the word denotes now. The word consider may belong on this list too, since it looks like con- plus sidus, sideris, a star, and may have begun as an augur's term for examining the sky. The dictionaries mark that derivation as probable rather than certain, and this is the right place to notice that a good etymology sometimes stops at probable.

The etymological fallacy

All of which leads to the error the whole lesson exists to name. The etymological fallacy is the belief that a word's origin gives its true meaning, so that a use departing from the origin is a mistake.

It is a fallacy because meaning is settled by use, and use moves. Nice comes from Latin nescius, not knowing, and went through silly, then fussy, then agreeable. Silly comes from Old English saelig, blessed, by way of innocent and feeble. Awful once meant full of awe. Naughty meant having nothing. Girl meant a child of either sex. Meat meant food of any kind, deer meant any animal, and starve meant simply to die. Every one of those words now means something its origin does not support, and every one of them is used correctly by hundreds of millions of people.

Bottom line: Etymology tells you where a word has been. It has no authority over where the word is.

How to check yourself

The method this course teaches produces hypotheses. Three warning signs mean you should verify before you commit.

1. The word is short and very common. The classical layer is mostly long and mostly technical. If it is a word a five-year-old uses, it is probably native and will not parse.
2. The parse needs an odd leap in meaning. If you have to argue for the sense as well as the shape, you are probably reconstructing a story rather than a word.
3. The result is too neat. Folk etymology exists precisely because tidy explanations are attractive. Sir Loin is a better story than surloigne, which is why it is the one people repeat.

Verification takes about twenty seconds in a dictionary that gives attested history rather than a guess. The Oxford English Dictionary is the reference standard where you have access; the Online Etymology Dictionary and Wiktionary are free and cite their sources.

Common misconceptions

• Etymology reveals a word's true meaning. It reveals the word's history. Nice, silly and awful all mean what speakers now use them to mean.
• Folk etymology is a mistake made by uneducated speakers. The s of island was inserted by scholars, and the ph of nephew by the same crowd. Learning is no protection.
• Female is male with a prefix. It is femella, a small woman, and the resemblance was manufactured in English.
• -ology always means the study of something. In anthology and trilogy the element means a gathering, from the other sense of legein.
• An etymology that fits perfectly is therefore right. Fitting perfectly is what a folk etymology does for a living.

What to remember

• Folk etymology reshapes an unfamiliar word to contain a familiar one, and the results are permanent: crayfish, belfry, island, female, hangnail, bridegroom, sirloin.
• The s in island was added by scholars to a native word that never had one, and the ph of nephew was added the same way.
• Greek legein means both to speak and to gather, so the -logy of biology and the -logy of anthology are different elements.
• The ending -ate makes verbs, adjectives, offices and chemical salts, and only context distinguishes them.
• Malaria, influenza, disaster, hysteria, lunatic and quarantine are accurate fossils of theories nobody holds.
• The etymological fallacy treats a word's history as a rule for its present use; meaning is settled by use, not by origin.
• Treat every parse as a hypothesis, and be most suspicious when it is tidiest.

Sources

1. Wikipedia. (n.d.). Folk etymology. Wikimedia Foundation. en.wikipedia.org
2. Harper, D. (n.d.). Island. Online Etymology Dictionary. etymonline.com
3. Harper, D. (n.d.). Belfry. Online Etymology Dictionary. etymonline.com
4. Wikipedia. (n.d.). Etymological fallacy. Wikimedia Foundation. en.wikipedia.org
5. Skeat, W. W. (1888). An etymological dictionary of the English language (2nd ed.). Clarendon Press.

Four words from an operative note

A surgical note reads: laparoscopic cholecystectomy, converted to open; incidental hepatic steatosis; no cholangitis. The patient it describes is entitled to know what happened to her, and every word in it comes apart.

• Laparoscopic: lapara, the soft part of the flank, plus skopein, to look at, plus -ic. Done by looking through the abdominal wall.
• Cholecystectomy: chole, bile, plus kystis, bladder, plus ektome, a cutting out. Removal of the gall bladder.
• Hepatic: hepar, hepatos, liver. Of the liver.
• Steatosis: stear, steatos, fat, plus -osis, an abnormal condition. Fatty change.
• Cholangitis: chole again, plus angeion, a vessel, plus -itis. Inflammation of the bile ducts.

So: the gall bladder was taken out through keyhole ports, the surgeon had to open the abdomen after all, the liver was fatty, and the bile ducts were not inflamed. Twelve Greek elements, one sentence, no jargon left.

Key idea: Medical terminology is not a vocabulary to be memorised. It is a grammar, and it was designed to be transparent to anyone who knows the parts.

Why every organ has two names

Anatomy was standardised in Latin, first informally and then formally, at the Basle congress of 1895 and in the Terminologia Anatomica that governs the field today. Clinical medicine and surgery, by a much older habit going back to the Hippocratic writers and to Galen, used Greek. The result is that most structures carry a Latin name for the part and a Greek name for what goes wrong with it.

The bottom row hides the word surgery itself. Greek cheirourgia is cheir, hand, plus ergon, work: hand-work. It went into Latin as chirurgia, into Old French as cirurgie, and into English as surgery, with almost nothing of the original spelling left. A surgeon is a manual worker, and the profession spent several centuries being reminded of it by physicians.

What matters here: The two systems are not interchangeable. A kidney has a renal artery and undergoes a nephrectomy, and no surgeon has ever performed a renectomy. The Latin base takes the anatomical adjective; the Greek base takes the disease and the operation.

The suffixes that do the clinical work

Three of those repay a second look because patients and students confuse them constantly. A tracheotomy is an incision into the windpipe. A tracheostomy is an opening into the windpipe that is kept open, usually with a tube. A tracheectomy would be removal of the windpipe, which is why nobody does one lightly. The three suffixes come from temnein, to cut, from the same verb with stoma, a mouth, and from ek plus temnein, to cut out. Getting them wrong on a consent form is not a spelling error.

A colostomy, on the same principle, is a mouth made in the colon and brought to the surface. The word describes the procedure exactly, and once you can read it the thing itself is much less frightening.

Four long words, taken apart

• Oesophagogastroduodenoscopy. oesophag- plus -o- plus gastr- plus -o- plus duoden- plus -o- plus -scopy. Looking at the gullet, the stomach and the duodenum. Note the joining vowels between the bases and their absence before the suffix, which is Lesson 5's rule doing its job. Duodenum is Latin: duodenum digitorum, of twelve fingers, because the segment is about twelve finger-breadths long.
• Otorhinolaryngology. ot- plus rhin- plus laryng- plus -logy. Ear, nose and throat, which is what everyone calls it.
• Electroencephalography. electr- plus en- plus cephal- plus -o- plus -graphy. Recording the electrical activity inside the head.
• Pancreaticoduodenectomy. pancreat- plus duoden- plus -ectomy: removal of the head of the pancreas together with the duodenum. Pancreas itself is pan plus kreas, all flesh, which is roughly what it looks like.

Three traps

ped- is three different elements. In pedal, pedestrian and pedicure it is Latin pes, pedis, foot. In paediatric, pedagogue and orthopaedic it is Greek pais, paidos, child. And a pedigree is neither: it is French pied de grue, a crane's foot, from the shape of the three-line mark used on a genealogical chart. American spelling, which drops the a of paediatric, removes the only visual clue that separates the first two.

hyster- is the womb. A hysterectomy removes the uterus. The emotional sense of hysteria comes from the ancient theory that the womb caused the condition, which Lesson 10 filed under fossils. The surgical word is not built on the emotional one.

my- and myel- are different. Greek mys, muscle, gives myalgia and myocardium. Greek myelos, marrow, gives myeloma and myelin and osteomyelitis. One letter apart, and one is a muscle while the other is the inside of a bone.

Common misconceptions

• -itis means infection. It means inflammation. Osteoarthritis, rheumatoid arthritis and contact dermatitis involve no organism at all.
• -oma means cancer. It means a mass. A lipoma is fat, a haematoma is a bruise and glaucoma is not a tumour, while melanoma and lymphoma are malignant. The suffix does not decide.
• The Latin and Greek names are interchangeable. Renal and nephric describe the same organ, but the operation is always a nephrectomy, and the two systems keep to their own jobs.
• A tracheotomy and a tracheostomy are the same procedure. One is an incision, the other leaves a maintained opening, and the suffix is the only thing that says so.
• Paediatrics is about feet. That is podiatry. The paed- is Greek for child, and the American spelling pediatrics removes the distinguishing letter.

Recap

• Clinical language is a grammar, not a glossary: laparoscopic cholecystectomy comes apart into five ordinary Greek ideas.
• Anatomy standardised on Latin and clinical practice on Greek, so most structures have two bases with different jobs.
• Surgery is Greek cheirourgia, hand-work, worn down through Latin and French past recognition.
• The suffix carries the clinical content: -itis inflammation, -osis condition, -oma mass, -algia pain, -megaly enlargement, -penia deficiency.
• -otomy cuts into, -ectomy cuts out, -ostomy makes a mouth, and the difference matters on a consent form.
• ped- means foot in Latin, child in Greek, and crane in pedigree; my- is muscle and myel- is marrow.
• The joining vowel rule from Lesson 5 governs these words exactly: it stays between bases and goes before a vowel suffix.

Sources

1. Wikipedia. (n.d.). Medical terminology. Wikimedia Foundation. en.wikipedia.org
2. Wikipedia. (n.d.). List of medical roots, suffixes and prefixes. Wikimedia Foundation. en.wikipedia.org
3. National Library of Medicine. (n.d.). MedlinePlus: Health topics. U.S. National Institutes of Health. medlineplus.gov
4. Federative International Programme for Anatomical Terminology. (2019). Terminologia anatomica (2nd ed.). Georg Thieme Verlag.
5. Dorland's illustrated medical dictionary (33rd ed.). (2019). Elsevier.

A trademark filed in 1899

In 1899 the Bayer company registered a name for acetylsalicylic acid: Aspirin. Three pieces. The a- is for acetyl. The -spir- is for Spiraea, the genus of meadowsweet from which salicylic acid had earlier been obtained. The -in is the ordinary chemical noun ending of the period, the same one in insulin, adrenalin and gelatin. Nobody chose the word for its sound; they chose it because it recorded the chemistry.

That is the habit of the whole field. Chemistry and pharmacology are the two trades that build the most new words from classical parts every year, and both do it under written rules. Read the rules and a bottle of pills becomes a legible document.

Key idea: In chemistry the informative part of a word is the ending, not the base. A generic drug name tells you what the drug does in its last two syllables.

Element names, sorted by what they record

One of those names is a scientific mistake that stuck. Lavoisier coined oxygen in 1787 from oxys, sharp or acid, plus the root of gignesthai, to be born, because he believed oxygen was the principle that made things acidic. Hydrochloric acid contains no oxygen at all, as Davy showed within thirty years. The name was never corrected.

The symbols that do not match the names

Eleven element symbols look wrong in English because they are abbreviations of the Latin name rather than the English one.

Plumbum earns a note. The Romans made pipes of lead, the tradesman who worked them was a plumbarius, and the word survives as plumber even though almost no plumbing has contained lead for a century. A plumb line is lead on a string, and to plumb the depths is to drop one.

The rule the naming committee now enforces

New elements are no longer named however their discoverers please. Under the IUPAC recommendations the ending must reflect the group: -ium for most of the table, -ine for group 17 with the halogens, and -on for group 18 with the noble gases. That is why element 117 is tennessine, matching chlorine and fluorine, and element 118 is oganesson, matching neon and argon.

Until an element is confirmed and named it carries a systematic placeholder built out of its atomic number, one digit at a time, using exactly the numeral elements of Lesson 6: nil, un, bi, tri, quad, pent, hex, sept, oct, enn. Element 117 was ununseptium and 118 was ununoctium.

The digit list is a deliberate mixture. un, bi, tri, quad, sept and oct are Latin; pent, hex and enn are Greek; nil is Latin for nothing. The committee chose whichever form gave a distinct initial letter, because the symbol is built from those initials: Uus, Uuo. Purity was never the point.

Numerals in a formula

Binary compounds take Greek numerals: carbon monoxide, carbon dioxide, dinitrogen tetroxide, phosphorus pentachloride, sulfur hexafluoride. Organic chemistry uses the same digits for chain length from five carbons upward, keeping four historical names below that: methane, ethane, propane, butane, then pentane, hexane, heptane, octane, nonane, decane.

The endings then carry the chemistry. -ane is a single bond, -ene a double bond, -yne a triple; -ol marks an alcohol, -al an aldehyde, -one a ketone, -oic acid a carboxylic acid, and -yl a group attached to something else. So ethanol is a two-carbon alcohol and ethene is a two-carbon chain with a double bond, and nobody had to be told either fact separately.

Three names for one drug

Every medicine has three names, and only the middle one is built to be read.

1. The chemical name, which describes the molecule exactly and is unusable in a clinic: paracetamol is N-(4-hydroxyphenyl)acetamide.
2. The generic name, assigned by the World Health Organization as an International Nonproprietary Name, or in the United States as a United States Adopted Name.
3. The brand name, owned by a company and chosen to be memorable.

The generic name is the one that carries information, because it ends in a stem assigned to a pharmacological class.

The antibody stem had a further layer. For years the syllable before -mab told you the source of the molecule: -ximab for a chimeric antibody, -zumab for a humanised one, -umab for a fully human one. In 2021 the World Health Organization abandoned -mab altogether for newly named products, because the source distinction had stopped being useful and the names were becoming unpronounceable. Four replacements were introduced: -tug, -bart, -mig and -ment. New antibodies now end in those.

Why this matters: A clinician meeting an unfamiliar generic name can place it in a class from the last syllable alone, without knowing anything else about the molecule. That is the same skill this course teaches on ordinary words, applied at the other end of the word.

Common misconceptions

• Oxygen is well named because acids contain oxygen. Lavoisier thought so in 1787 and was wrong. Hydrochloric acid has none, and the name is a fossil of a discarded theory.
• Element symbols abbreviate the English name. Fe, Pb, Au, Ag, Cu, Sn, Hg, Sb, K and Na abbreviate the Latin one.
• Generic drug names are meaningless strings. The ending is assigned, not invented, and it names the class.
• Aspirin has always been a generic word. It was a Bayer trademark. Bayer lost it in several countries under the Treaty of Versailles in 1919, and it remains a trademark in Germany and Canada to this day.
• The systematic element names are pure Latin. They mix Latin and Greek digits deliberately, so that every digit has a distinct initial letter for the symbol.

The short version

• Aspirin is acetyl plus Spiraea plus the chemical ending -in, and it was a trademark before it was a word.
• Element names record properties, colours, smells, places, people and myths, and oxygen records a mistake.
• Ten symbols abbreviate Latin rather than English, which is why lead is Pb and the tradesman is a plumber.
• IUPAC now requires -ium, -ine or -on according to the group, giving tennessine and oganesson.
• Placeholder names spell the atomic number in mixed Latin and Greek digits, so 117 was ununseptium.
• Organic endings carry the chemistry: -ane, -ene, -yne, -ol, -al, -one and -oic acid.
• A generic drug name is classified by its stem, and the antibody stem -mab was replaced in 2021 by -tug, -bart, -mig and -ment.

Sources

1. International Union of Pure and Applied Chemistry. (n.d.). Periodic table of elements. IUPAC. iupac.org
2. World Health Organization. (n.d.). International Nonproprietary Names. WHO. who.int
3. American Medical Association. (n.d.). United States Adopted Names approved stems. AMA. ama-assn.org
4. Harper, D. (n.d.). Aspirin. Online Etymology Dictionary. etymonline.com
5. Connelly, N. G., Damhus, T., Hartshorn, R. M., & Hutton, A. T. (Eds.). (2005). Nomenclature of inorganic chemistry: IUPAC recommendations 2005. RSC Publishing.

A barrel of flour, Liverpool, 1863

A man was walking along a street in Liverpool when a barrel of flour fell out of a first-floor window of Boadle's shop and knocked him down. He could not say who had been moving the barrel, or how it came loose, or what anyone had failed to do. On the ordinary rule he had no case, because a plaintiff must prove the defendant was careless and he could prove nothing at all.

The Court of Exchequer decided for him anyway. Chief Baron Pollock said that there are certain cases of which it may be said res ipsa loquitur. The thing speaks for itself. Barrels in properly run warehouses do not fall out of windows, so the fall is itself evidence that somebody was careless, and it is now for the defendant to explain. Byrne v Boadle is cited in negligence cases to this day, and it is cited by that Latin phrase, because the phrase is the name of the doctrine.

Key idea: Legal Latin is not decoration. A phrase such as res ipsa loquitur is a compressed technical term whose content is a hundred and sixty years of decided cases. That is also the strongest argument against translating it.

Why two dead languages, and not one

English law accumulated its vocabulary in layers, and each layer arrived in a different language.

• Latin came from the church and the universities. Canon law and Roman law were taught and written in Latin, court records were kept in Latin, and writs began with Latin words that became their names.
• Law French came from the Norman conquest. French was the spoken language of the English courts for three centuries. The Statute of Pleading of 1362 required cases to be pleaded in English, on the ground that litigants could not understand the proceedings, while the records continued in Latin. Law French survived anyway, in reports and textbooks, long after it had stopped being anybody's French: by the fifteenth century it was a technical jargon with English syntax and invented words that no visitor from Paris could have followed.
• English finally took over by statute. An Act of 1730, in force from 1733, required that proceedings be in English, which ended Latin and Law French as languages of record and left them as vocabulary.

The French layer

The French layer also left a syntax. French puts the adjective after the noun, and legal English kept the order in fixed phrases: attorney general, court martial, fee simple, malice aforethought, letters patent, notary public, heir apparent, body politic, condition precedent. That is why the plural is attorneys general and courts martial, with the -s on the noun where it belongs, and why writing attorney generals still sounds wrong to lawyers.

A third habit came from drafting in two languages at once. A clerk who was not sure his readers had the same vocabulary wrote both words, and the pairs fossilised: last will and testament, English beside Latin; goods and chattels, English beside French; breaking and entering; aid and abet; free and clear; fit and proper; cease and desist; give, devise and bequeath. Modern drafters cut these where they can, and courts occasionally have to decide whether a pair was two ideas or one.

The Latin phrases, and what each actually does

Three of these repay a closer look. Habeas corpus is not a right to a body or a demand to see one; it is the second and third words of a mediaeval writ, and the whole document is named after its opening, exactly as a papal encyclical is. Affidavit is a whole Latin sentence, he has declared on oath, doing duty as an English noun. And versus is the participle of vertere, to turn, so a case name says that one party is turned against the other.

The core of it: Most legal Latin is either the first words of a document or a compressed rule. Neither kind is ornamental, which is why replacing them is harder than it looks.

The argument about getting rid of it

England and Wales tried. Lord Woolf's review of civil justice produced the Civil Procedure Rules, in force from April 1999, which replaced a list of terms with plain English: plaintiff became claimant, writ became claim form, in camera became in private, ex parte became without notice, subpoena became witness summons, and the Mareva injunction and Anton Piller order became the freezing injunction and the search order.

The case for was straightforward. Litigants in person have to read these documents, and a word that has to be explained before it can be used is a barrier disguised as precision. The case against was that centuries of decided cases attach to the old terms, and that a new word invites new argument about whether it means the same thing.

Both turned out to be partly right. The everyday procedural words changed and nobody misses them. The doctrinal phrases did not: res ipsa loquitur, mens rea, ultra vires and stare decisis are all still in daily use, in England and in the United States, where plaintiff also survives untouched. Words that named a procedure were replaceable; words that named a rule were not.

Common misconceptions

• A tort is a crime. It is a civil wrong, actionable by the injured party. The same act can be both, which is why a defendant may be acquitted and still lose a damages claim.
• Law French is French. By the fifteenth century it was a professional jargon with English word order and invented vocabulary, unintelligible to French speakers.
• Habeas corpus means the right to see the body. It is the opening of a writ commanding a gaoler to produce a prisoner and justify holding him.
• Legal Latin is ornamental. Some of it is; most of it is compression, and a doctrine that took a hundred cases to define is not easily renamed.
• Attorney means lawyer. It means an appointed representative, which is why a power of attorney can be given to someone with no legal training at all.

Putting it together

• Byrne v Boadle in 1863 gave the negligence doctrine that is still known by its Latin name, res ipsa loquitur.
• Latin entered legal English through the church, the universities and the written record; French entered through the Norman courts and stayed as a jargon.
• The 1362 statute put pleading into English and the Act of 1730 finished the job, leaving both languages as vocabulary rather than as media.
• French left word order behind: attorney general, court martial, malice aforethought, and their plurals.
• Bilingual drafting left doubled phrases such as will and testament and goods and chattels.
• Most Latin tags are either the opening words of a document, as with habeas corpus and subpoena, or a compressed rule, as with stare decisis and ultra vires.
• The 1999 reforms replaced the procedural words and left the doctrinal ones alone, which is a fair test of which kind was doing work.

Sources

1. Legal Information Institute. (n.d.). Res ipsa loquitur. Cornell Law School. law.cornell.edu
2. Legal Information Institute. (n.d.). Habeas corpus. Cornell Law School. law.cornell.edu
3. Wikipedia. (n.d.). Law French. Wikimedia Foundation. en.wikipedia.org
4. Mellinkoff, D. (1963). The language of the law. Little, Brown.
5. Garner, B. A. (Ed.). (2019). Black's law dictionary (11th ed.). Thomson Reuters.

A worm on a whale, 125 metres down

In 2005 a team working on a minke whale carcass on the floor of a Swedish fjord described a new animal and gave it the name Osedax mucofloris. Take it apart. Os, ossis is Latin for bone, and -edax is from edere, to eat, the element also in edible: bone-devourer. Mucus is Latin for what it looks like, and flos, floris is a flower. The English common name the describers used was the bone-eating snot-flower worm, which is a translation, not a joke.

Nothing about that name is ancient. It was manufactured in the twenty-first century out of parts two thousand years old, under rules that require exactly this, and the reason is that every biologist on earth has to be able to read it.

Key idea: A scientific name is not a label to be looked up. It is a small Latin phrase, and it obeys Latin grammar, which is why it can be read by someone who has never seen the organism.

What a binomial is made of

Carl Linnaeus fixed the system in two books: Species Plantarum in 1753, which is the formal starting point for plant names, and the tenth edition of Systema Naturae in 1758, which is the starting point for animal names. Before him a species was identified by a descriptive Latin phrase that could run to a dozen words. Linnaeus kept the phrase for the description and reduced the name to two words.

• The genus: a Latin or Latinised noun in the nominative singular, capitalised, italicised. Panthera, Quercus, Escherichia, Osedax.
• The specific epithet: lowercase, italicised, and grammatically dependent on the genus. leo, alba, coli, mucofloris.
• Often an author and date after the name, unitalicised: Homo sapiens Linnaeus, 1758.

Strictly, the species name is both words together. The second word alone is the specific epithet in botany and the specific name in zoology, and it means nothing on its own: alba occurs in dozens of genera.

Three kinds of epithet, and how to tell them apart

The agreement rule is where an outsider first stumbles, and it has a classical trap in it. Latin tree names are feminine whatever their endings look like, so Quercus, the oak, is feminine and the white oak is Quercus alba, not Quercus albus. Acer, the maple, is neuter, so the red maple is Acer rubrum. Pinus, Fagus, Ulmus and Populus are all feminine for the same reason. A botanist who writes Quercus albus has made a grammatical error, and the code says so.

Personal epithets take the genitive and the ending records the person's number and gender: -i for one man, -ae for one woman, -orum and -arum for groups. Rafflesia arnoldii honours Joseph Arnold; the wasp Aha ha honours nothing but its describer's sense of humour.

Worth holding on to: If the epithet ends in -us, -a or -um, look at the genus and check the gender. If it ends in -i or -ae, somebody is being commemorated. If it is a bare noun such as leo or tigris, no agreement is required.

The epithets you will meet again and again

Latin in form, not in origin

The codes do not require that a name be a Latin word. They require that it be Latin in form, that is, treated as Latin for the purposes of grammar and declension. That leaves the door open to everything.

• Ginkgo is Japanese, and it is misspelled. Engelbert Kaempfer transliterated the Japanese reading in the early eighteenth century, the form ginkgo appeared in print, and Linnaeus used it. Under the rule of priority the misspelling is now the correct name of the tree.
• Gorilla comes from a Greek account of a Carthaginian voyage, describing hairy people on the West African coast, and was applied to the ape in 1847.
• Tyrannosaurus rex is Greek tyrannos plus Greek sauros, then Latin rex. It is a hybrid of exactly the kind Lesson 16 examines, and it is entirely legal.
• Wollemia nobilis, a conifer found alive in 1994 in a canyon in New South Wales, takes its genus from the Wollemi National Park and its epithet from David Noble, who found it, by way of a Latin adjective that also means noble.

Describers occasionally play. Agra vation is a beetle. Ba humbugi is a Fijian land snail. Han solo is a trilobite. Aha ha is a wasp. Spongiforma squarepantsii is a fungus. All are valid names, because validity turns on publication and form, not on dignity.

Why a wrong name cannot be fixed

The governing principle of the zoological code and its botanical counterpart is priority: the earliest validly published name wins, and it wins even when it is wrong.

Basilosaurus means king lizard. Richard Harlan described the animal in 1834 from vertebrae found in Louisiana and took it for a marine reptile. It is a whale. The error was recognised within a few years, and the name stands, because changing names when opinions change would destroy the stability the whole system exists to provide. Every biologist who writes about Basilosaurus knows it is a whale, and every one of them calls it a king lizard.

Two other conventions follow from stability. Tautonyms, where the genus and epithet are identical, are permitted in zoology and produce Bison bison, Gorilla gorilla and Vulpes vulpes; botany forbids them. And ranks above genus take standard endings, so a family of animals ends in -idae, as Felidae and Hominidae do, while a family of plants ends in -aceae, as Rosaceae does.

Common misconceptions

• A scientific name must be made of Latin words. It must be Latin in form. Ginkgo is Japanese, Gorilla is from a Greek travel narrative, and both are correct names.
• Tyrannosaurus rex is bad practice because it mixes Greek and Latin. Hybrid names are legal and common. The codes regulate form and publication, not purity.
• Quercus is masculine because it ends in -us. Classical tree names are feminine regardless of ending, which is why it is Quercus alba.
• A name can be corrected once the science changes. Priority protects the name. Basilosaurus is a whale called king lizard, permanently.
• australis means Australian. It means southern. Australia was named from the same root, for the supposed southern land, not the other way round.

Where this leaves us

• A binomial is a two-word Latin phrase: a capitalised genus noun and a lowercase epithet that depends on it grammatically.
• Epithets are adjectives that must agree in gender, nouns in apposition that do not, or genitives naming a person or a host.
• Latin tree names are feminine, so it is Quercus alba, Pinus nigra and Acer rubrum with a neuter genus.
• About thirty epithets, from vulgaris and officinalis to repens and foetida, cover an enormous share of the names you will meet.
• The codes require Latin form, not Latin origin, which is why Ginkgo, Gorilla and Tyrannosaurus rex are all valid.
• Priority makes names stable at the cost of accuracy: Basilosaurus is a whale and always will be a king lizard.
• Family endings are fixed, -idae for animals and -aceae for plants, so the rank is legible from the word.

Sources

1. Wikipedia. (n.d.). Binomial nomenclature. Wikimedia Foundation. en.wikipedia.org
2. International Commission on Zoological Nomenclature. (n.d.). The International Code of Zoological Nomenclature. ICZN. iczn.org
3. Wikipedia. (n.d.). Osedax. Wikimedia Foundation. en.wikipedia.org
4. Integrated Taxonomic Information System. (n.d.). ITIS standard report search. U.S. Department of Agriculture and partners. itis.gov
5. Stearn, W. T. (1992). Botanical Latin (4th ed.). David and Charles.

A word invented in 1958 to frighten people

In 1958 Michael Young published a short book called The Rise of the Meritocracy, 1870-2033. It is a sociological report written from the year 2033 by a narrator who approves of everything he describes: an England sorted by tested intelligence, in which the clever are promoted out of their families and everybody else is told, with evidence, that they deserve what they got. A footnote records that the author of the report was killed in the riots of May 2033. The book is a satire, and meritocracy is its monster.

In 2001 Young, by then Lord Young of Dartington, published a piece in the Guardian under the title Down with Meritocracy, complaining that Tony Blair's Labour Party kept using his word as a compliment. He had built it as an insult.

So what?: The -cracy words are not neutral labels for kinds of government. A good many of them were coined by somebody who disapproved of the thing being named, and the ending carries that history with it.

Two Greek nouns, and they are not the same noun

Everything in this lesson comes out of two Greek words.

• kratos, strength, might, the power to prevail. kratia gives English -cracy, the agent noun krates gives -crat, and the adjective gives -cratic. The same element sits inside Hippocrates, which is hippos plus kratos, horse-power.
• arche, which means both a beginning and a rule. The verb archein means to be first, and being first covered both starting a thing and commanding it. That double sense gives -archy and -arch, and also archaic and archetype.

Hold the difference, because it predicts most of what follows. kratos is power held by somebody; arche is the position of being first. So the -archy words mostly count the rulers, and the -cracy words mostly name the group that holds the power.

That is a tendency with named exceptions, not a law, and it is still the fastest way to guess what an unfamiliar word of either family is claiming.

The -cracy family, with dates where we have them

The -archy family

Making the person and the adjective

Each family has its own derived forms, and they do not swap.

A monarchist is not a monarch: -ist makes a partisan, the same contrast that runs through royalist and federalist. And anarch is real but literary, Pope's great Anarch letting the curtain fall at the end of the Dunciad, so the person attached to anarchy in ordinary English is the anarchist.

In short: -cracy takes -crat and -cratic. -archy takes -arch, and then -archic or -archal. The -ist forms name supporters of a system rather than holders of an office.

The other arche, and the arch- on the front

arche meant a beginning before it meant a rule, and English has both senses.

• The beginning sense: archaic and archaeology, from archaios, ancient; archetype, arche plus typos, the first model struck; and in philosophy the arche is whatever a thinker makes the first principle of things, water for Thales.
• The rule sense: archive, which looks like the beginning sense and is not. It comes through Latin archivum from Greek arkheion, the building where the archons kept the public records.

Then there is arch- on the front of a word, from Greek archi-, chief: archbishop, archangel, archduke, archrival, archfiend. architect is archi plus tekton, a builder, hence the chief builder; archipelago is archi plus pelagos, chief sea, which named the Aegean long before it meant a group of islands.

A pronunciation rule hides in that list, and it is nearly regular. Before a consonant arch- sounds like the arch of a bridge: archbishop, archduke, archrival. Before a vowel it sounds like ark: archangel, architect, archipelago, archaeology. An archbishop and an archangel carry the same prefix and say it differently.

What Latin supplied instead

Greek gave English its words for kinds of government. Latin gave it the words for offices and for the people holding them, and the literal sense is usually sharper than the modern one.

The pair to notice is magister and minister, because the Latin words are exact opposites and English kept both without anyone remarking on it. magis is more, minus is less. The master is the greater man and the minister the lesser one, the servant. A cabinet minister called a servant of the Crown is not being paid a compliment; that is the word.

Governor is the other. Latin took kybernan, to steer a ship, from Greek and made it gubernare, which gave English governor. In 1948 Norbert Wiener went back past the Latin and built cybernetics on kybernetes, the steersman, so governor and cyber- are one word arriving by two doors: the shape of the doublets in Lesson 7.

Common misconceptions

• Anarchy means chaos. It means an- plus arche, without a ruler. Whether that produces disorder is the political question, and every anarchist since Proudhon has answered no.
• -cracy and -archy are interchangeable. They rest on different nouns: kratos is power held, arche is the position of being first. Hence the counting words in one family and the naming words in the other.
• Hierarchy simply means a ranking. It was made about AD 500 for the ranks of angels and then of clergy, on hieros, sacred. The organisation-chart sense is a late extension.
• Bureaucracy is a Greek word. Half of it is French: bureau is a desk, and the compound was assembled in French about 1750, which makes it the same kind of hybrid as meritocracy.
• Meritocracy has always been praise. Michael Young wrote it into a dystopia in 1958 and complained in 2001 that it had been turned into a slogan.

The takeaway

• Two Greek nouns run the vocabulary: kratos, power held, giving -cracy, -crat and -cratic; arche, being first, giving -archy, -arch and -archic.
• The -archy words mostly count the rulers and the -cracy words mostly name the group with the power, with autocracy, hierarchy and patriarchy as the named exceptions.
• Members of both families are datable inventions: theokratia about AD 94, hierarchy about AD 500, bureaucratie about 1750, technocracy in 1919, polyarchy in 1971.
• arche also means beginning, which puts archaic and archetype in the family and leaves archive, against appearances, in the rule sense.
• arch- on the front means chief, said like arch before a consonant and like ark before a vowel.
• Latin supplied the offices rather than the systems: the minister is the lesser man, the candidate the whitened one, the governor the steersman.

Sources

1. Wikipedia. (n.d.). Meritocracy. Wikimedia Foundation. en.wikipedia.org
2. Wikipedia. (n.d.). Hierarchy. Wikimedia Foundation. en.wikipedia.org
3. Online Etymology Dictionary. (n.d.). -cracy. Harper, D. etymonline.com
4. Wiktionary. (n.d.). -archy. Wikimedia Foundation. en.wiktionary.org
5. Young, M. (1958). The rise of the meritocracy, 1870-2033. Thames and Hudson.

A fringe like a crown, photographed in 1967

June Almeida, working at St Thomas's Hospital in London, put a sample from a common cold on an electron microscope grid and produced pictures of particles about a ten-thousandth of a millimetre across, each one ringed by a fringe of club-shaped projections. She and David Tyrrell had been chasing a cold virus, strain B814, that would not grow in the usual cultures. What she photographed did not look like influenza and did not look like anything else in the catalogue.

The fringe reminded Almeida and Tyrrell of the corona, the halo of light around the sun in an eclipse, and the name they coined went into print in 1968, in a short note by an informal group of virologists in Nature. Corona is Latin for a garland or a crown, itself borrowed from Greek korone, and the family became the coronaviruses. Virus is Latin too, for poison or slime, taken straight and left almost unaltered.

Fifty-two years later, on 11 February 2020, a taxonomy committee named a new member SARS-CoV-2 and the World Health Organization named the illness it causes COVID-19. The second of those is not a compound at all. It is an acronym: COronaVIrus Disease, plus the year. One word in the pair was assembled out of Latin; the other was cut out of English initials and only looks classical.

Why this matters: The elements you have spent this course learning are not a closed inventory of old words. They are a live workshop, and things come out of it every year. But not everything that looks classical was built that way, and telling the two apart is the last skill this course owes you.

Omicron, and the letters that are compounds

On 31 May 2021 the WHO announced that variants of SARS-CoV-2 would be labelled with Greek letters, so that nobody would have to say the Kent variant or the South African variant. Alpha, beta, gamma and delta went by. On 26 November 2021 the organisation reached the fifteenth letter and named the Omicron variant, having skipped the two letters in front of it: nu, because spoken aloud it is indistinguishable from new, and xi, because it is a common surname.

Here is what most of the reporting missed. Omicron is itself a compound, and so are three of its neighbours.

The small letters carry English words too. Iota, the smallest letter, gives not one iota and also, through Latin, the word jot. Delta gave the Nile delta its name because the land at the river's mouth is a triangle.

Meme, a word built short on purpose

In the last chapter of The Selfish Gene, in 1976, Richard Dawkins wanted a noun for a unit of culture that copies itself. He took the Greek mimeme, a thing imitated, from mimeisthai, to imitate, the same root as mime and mimic, and then cut it down. He wanted a monosyllable that would sound like gene, so mimeme lost its middle and became meme.

Two things about that are worth noticing. First, he deliberately damaged the Greek to get the shape he wanted, and nobody has ever complained. Second, the word has since drifted a very long way from what he meant by it, until it now mostly denotes a picture of a cat with writing on it. If you told a room of eighteen-year-olds that meme really means a self-replicating unit of culture, you would be committing the etymological fallacy from Lesson 10 in its purest form. The word means what it is used to mean.

Hybrids, and the rule everyone breaks

A hybrid word takes its parts from two languages. Purists have objected to them for two centuries, and the objection has lost every round.

There is a well-worn quip, usually attributed to C. P. Scott of the Manchester Guardian, that television is half Greek and half Latin and no good can come of it. Whoever said it, the prediction failed on the linguistic point. Hybrids are now the ordinary case, because a coiner reaches for whichever element is already familiar to the audience, and because the -o- that joins Greek compounds has quietly become the default joint for new words of any parentage.

Bottom line: Purity of stock is not a rule of English word formation and never really was. What matters is whether the parts are recognisable, which is a fact about readers rather than about Greek.

Elements that were never elements

Some very productive modern endings were cut out of words that had no such piece in them. They are worth knowing because they will defeat a parse.

• -athon, from marathon. Marathon is a place in Attica, a deme whose name has nothing inside it. English sliced it after the first two letters and produced telethon, walkathon and readathon.
• -gate, from Watergate, the name of a building in Washington. Every scandal since has been able to take the ending.
• -holic, from alcoholic, which divides as alcohol plus -ic. Alcohol is Arabic, al-kuhl, the powdered antimony used as eye paint. Workaholic and chocoholic are built on a cut that goes through the middle of an Arabic noun.
• -ome and -omics. Hans Winkler coined Genom in German in 1920 out of gene and chromosome. The -ome he ended up with was an accident of the second word, not a Greek suffix. In 1986 a geneticist proposed genomics as the name of a new journal, and the pattern was away: proteome in 1994, then transcriptome, metabolomics, connectome, microbiome.

That last one has a trap in it. Greek does have a suffix -oma, and it means a swelling or a mass: carcinoma, sarcoma, haematoma, glaucoma, and the -oma of Lesson 11. It is not the -ome of genome, despite the resemblance, and a microbiome is not a tumour of microbes.

Coinages you can put a date on

Anthropocene shows how a productive series works. Charles Lyell built the names of the Tertiary epochs on kainos, recent, with a quantity word in front: Eocene the dawn-recent, Miocene the less-recent, Pliocene the more-recent, Pleistocene the most-recent, Holocene the wholly-recent. Crutzen and Stoermer simply took the next slot and put anthropos in it. Nobody had to explain the ending.

Where the Latin went: trademarks

Latin left public life and reappeared on packaging. A brand name has to be short, pronounceable in many languages and registrable, and a dead language is an unusually convenient quarry.

Generic drug names are the same trick under stricter rules, and Lesson 12 worked through the stem system that governs them. What the trademarks add is that the meaning is often decorative: nobody buying a Volvo is thinking about rolling, and the word survives because it is short and says nothing rude in any major language.

Common misconceptions

• COVID is a compound like coronavirus. It is an acronym of coronavirus disease plus a year. Coronavirus is a Latin compound; COVID-19 only looks like one.
• Hybrid words are mistakes. Television, automobile, sociology, neonatal and bureaucracy are all hybrids, and they are the normal case rather than the exception.
• Marathon contains an element -athon. It is a place name in Attica with no parts. The ending was cut out of it in the twentieth century and lent to telethon.
• The -ome of genome is the -oma of carcinoma. They are unrelated. Greek -oma is a swelling; the -ome of genome fell off the end of chromosome in 1920.
• Omicron was chosen for its meaning. It was the next letter available after nu and xi were passed over, and it happens to mean little o.
• Classical word-building is finished. Every entry in the dated table above is younger than the steam engine, and several are younger than the reader.

Looking back

• Almeida and Tyrrell named the coronaviruses for the crown of projections she had photographed, and the word went into print in Nature in 1968; corona is Latin for a garland.
• COVID-19 is an acronym, not a compound, which is the cleanest illustration of the difference in the language.
• Omicron, omega, epsilon and upsilon are compound letter names: little o, big o, bare e, bare u.
• Dawkins clipped mimeme to meme in 1976 to get a monosyllable that rhymed with gene, and the sense has since moved a long way from his.
• Hybrids of Greek and Latin are ordinary, not erroneous, and the Greek joining vowel -o- has become the default joint for new words of any origin.
• -athon, -gate, -holic and the -ome of genome were all cut out of words that never contained them, and will defeat a parse.
• Datable coinages run from van Helmont's gas about 1640 to the Anthropocene in 2000, and the Latin that left public life turned up again on packaging.

Sources

1. Wikipedia. (n.d.). Coronavirus. Wikimedia Foundation. en.wikipedia.org
2. World Health Organization. (n.d.). Tracking SARS-CoV-2 variants. WHO. who.int
3. Wikipedia. (n.d.). Anthropocene. Wikimedia Foundation. en.wikipedia.org
4. Online Etymology Dictionary. (n.d.). meme. Harper, D. etymonline.com
5. Dawkins, R. (1976). The selfish gene. Oxford University Press.

A council that met once a week

From 1854 until 2000 the University of Oxford ran its business through a body called the Hebdomadal Council. Nobody who sat on it thought the name odd, and nobody outside it knew what the word meant. You can get most of the way there now without help. Greek hepta is seven; the derived noun hebdomas, hebdomados is a group of seven, hence a week; -al makes an adjective. Hebdomadal means weekly, and the council met once a week.

That is the whole of this course in four sentences, and this lesson is nothing but the same thing done twenty more times, slowly at first and then at speed. The words below are real, current and, unless you have an unusual reading history, unfamiliar. None of them will be defined before it is parsed.

The procedure, in seven steps

1. Strip the grammar. Take off plural -s, past -ed, participial -ing and adverbial -ly. They are English and carry no classical information.
2. Take off the outermost suffix and note what part of speech it makes: -ous, -al, -ar, -ic and -ine make adjectives; -ion, -ation, -y, -sis and -ism make nouns; -ate and -ise make verbs.
3. Look for a prefix. If the word opens with a doubled consonant or an awkward cluster, suspect an assimilated prefix and restore it: ad-, in-, con-, sub-, ob-, ex-, dis-, as in Lesson 2.
4. Find the base or bases, and check the joint. A connecting -o- says Greek; a connecting -i- says Latin. A word with two bases usually reads back to front: the last base is the head.
5. Check the spelling tells. ph, th, ch, rh, y between consonants, ae or oe, and initial ps-, pn-, pt- or mn- all point to Greek. Lesson 5 has the full table.
6. Assemble a literal gloss, suffix first, then the head base, then whatever modifies it.
7. Test the gloss against a real sentence, then check a dictionary. The parse narrows the possibilities to a handful. It does not choose among them, and it cannot tell you where a meaning has drifted.

Remember: Steps one to six are mechanical and you will get faster at them. Step seven is not optional, and the rest of this lesson is largely about why.

Five words, worked slowly

1. anfractuous. Strip -ous, an adjective ending: anfractu-. That is the Latin noun anfractus. The base is fract-, which you met in Lesson 3 as the fourth principal part of frangere, to break, and again in Lesson 7 in frail and fragile. What is an-? Not the Greek negative, which does not attach to Latin verbs. It is ambi-, around, which loses its b before a consonant. Literally: broken around, hence bent round on itself. Anfractuous means winding, full of turns. Anatomists still call the folds of the brain its anfractuosities.

2. autochthonous. The cluster chth is a spelling tell that admits no argument: this is Greek, and Latin has nothing like it. Strip -ous. auto- is autos, self, joined with -o-, which confirms Greek. chthon is the earth, the ground underfoot, as against ge, the earth as a whole. Literally: from the earth itself. It means native to the place where it is found, and in medicine an autochthonous infection is one caught locally rather than imported.

3. saxicolous. Strip -ous. The joint is -i-, so this is Latin. saxum is a rock; -col- is colere, to inhabit or to till, which also gives colony and cultivate. Literally: rock-inhabiting. It is the standard word for a lichen or plant that grows on stone. The family is large and regular: terricolous on soil, cavernicolous in caves, arboricolous on trees, and agricola, the Latin farmer, who is a field-dweller.

4. omphaloskepsis. The -o- joint and the ending -sis both say Greek. omphalos is the navel, and the stone at Delphi that the Greeks called the navel of the world. skepsis is examination or inquiry, from skeptesthai, to look at, which also gives sceptic. Literally: navel-examination. It means exactly what navel-gazing means in English, and it was built as a joke.

5. fissiparous. The -i- joint says Latin. fissus is the participle of findere, to split, and gives fission and fissure. -parous is parere, to bring forth, the element in viviparous, bearing live young, and oviparous, laying eggs. Literally: split-bearing. In biology it describes an organism that reproduces by dividing. Applied to a political party it means one that keeps splitting, which is a metaphor a biologist would recognise.

Eight words, at speed

The upshot: Eight words, none of them common, and the parse landed close enough on every one that a dictionary would only be confirming you. That is the good case. The next seven are the honest ones.

Seven words where the parse only gets you halfway

14. latitudinarian. latus is wide; the noun is latitudo, and Lesson 4 told you to take the stem from the genitive, latitudinis, which gives latitudin-. Add -arian, one who favours. Literally: one who allows breadth. The parse is correct and would not tell you that the word belongs to seventeenth-century English church history, where a latitudinarian was a clergyman relaxed about doctrine and ceremony.

15. tergiversation. tergum is the back, versare is to turn. Literally: back-turning. Now choose: does it mean evasion and shuffling, or desertion of a cause? It means both, and the parts cannot tell you which is meant in a given sentence. Only the sentence can.

16. psephology. The initial ps- is a Greek tell. psephos is a pebble; -logy is the study of. Literally: the study of pebbles. It means the study of elections and voting patterns. W. F. R. Hardie built the word in Oxford in 1948, for his friend R. B. McCallum, who had asked for a name for the subject, and it reached print in 1952. The gap between gloss and meaning closes only if you happen to know that Athenians voted by dropping pebbles into urns. No amount of care with the morphology gets you there.

17. paludal. palus is a marsh, genitive paludis, stem palud-. Literally: of the marsh. Correct, and still not enough, because paludal fever is malaria. The link is a discarded theory: the disease was blamed on the air over marshes. Malaria itself proves the point twice over, since it is not classical at all but Italian mala aria, bad air.

18. succedaneum. Two doubled consonants and an odd opening: suspect assimilation. sub- becomes suc- before c, and the base is cedere, to go or yield, which gives succeed, recede, precede and concede. Literally: a thing that goes in underneath. It means a substitute, usually a drug substituted for another. It also means the swelling on a newborn's scalp after delivery, the caput succedaneum, and nothing in the parts would ever have suggested that.

19. hypnopompic. hypnos is sleep; pompe is a sending, from pempein, to send, and is also the ancestor of pomp. Literally: sleep-sending-away. It describes the state of coming out of sleep, and it has a twin, hypnagogic, hypnos plus agogos, leading, for the state of falling into sleep. Frederic Myers made the pair in the 1890s. The parse gives you the two words only once somebody tells you that pompe sends away and agogos leads in.

20. inspissated. Strip -ed and -ate. spissus is Latin for thick or dense. And in-? Here is the trap this course has been walking towards. in- does two entirely different jobs: negative, as in invisible and inarticulate, and intensive or directional, as in inflame, inscribe and ingrain. Nothing in the word marks which. inspissated means thickened, not un-thickened, and the only way to know is to ask what would make sense, then check. Lesson 10 called the general case a false friend; this is the same trap inside a single prefix.

When the procedure fails outright

Three kinds of word will defeat the method entirely, and recognising them quickly is worth as much as any parse.

• Native English. understand is under plus stand, forgive is for- plus give, gospel is god plus spell, good news, and lord was hlaford, the loaf-warden. Long, opaque and completely un-classical.
• Loans in disguise. alcohol, algebra and alkali all begin with the Arabic article al-, not a Latin prefix. algorithm is a man's name, al-Khwarizmi. admiral is the worst of them: Arabic amir al-, commander of the, reshaped in Europe by association with Latin admirari, to wonder at, which is where the intrusive d came from. Parse admiral as ad- plus mir- and you will get a confident, wrong answer.
• Words whose sense has moved. The parts of decimate, nice and prestigious are all recoverable and all misleading. Lesson 10 set out why the origin does not govern the current meaning, and the twenty words above included three, psephology, paludal and succedaneum, where the same gap opens.

Common misconceptions

• A correct parse gives the meaning. It gives a literal gloss. psephology really is the study of pebbles and really does mean the study of elections, and only a fact about Athenian voting joins the two.
• in- always means not. It is negative in invisible and intensive in inflame and inspissated. The prefix is ambiguous and the word will not tell you which.
• ante- and anti- are spellings of one prefix. ante is Latin for before, as in antediluvian and antebellum; anti is Greek for against, as in antibiotic. They are unrelated.
• Any long word has classical parts. understand, admiral and algorithm are all long and none of them will parse.
• Using a dictionary means the method failed. The method turns a blank into a shortlist of two or three readings. Checking which one is right is the last step, not an admission of defeat.

What to carry forward

• The procedure is fixed: strip the grammar, take off the suffix, restore any assimilated prefix, find the bases, read the joint and the spelling for the language, gloss back to front, then check.
• The joining vowel remains the fastest single test: -o- for Greek, as in omphaloskepsis and anemochory, -i- for Latin, as in saxicolous and fissiparous.
• Spelling tells are decisive when they appear: chth, rh-, ps-, ph and y between consonants are Greek and nothing else.
• A parse yields a literal gloss, not a definition, and three things can still separate the two: a lost theory, as in paludal; a lost practice, as in psephology; and a technical narrowing, as in succedaneum.
• in- is ambiguous between negative and intensive, so inspissated is thickened, not un-thickened.
• Native English compounds and Arabic loans wearing Latin spelling will not parse, and admiral is the standing warning.
• Twenty unfamiliar words went through the same seven steps, and the steps did not change for any of them. That is the point of having a procedure.

Sources

1. Wiktionary. (n.d.). psephology. Wikimedia Foundation. en.wiktionary.org
2. Wiktionary. (n.d.). anfractuous. Wikimedia Foundation. en.wiktionary.org
3. Online Etymology Dictionary. (n.d.). antediluvian. Harper, D. etymonline.com
4. Wikipedia. (n.d.). List of Greek and Latin roots in English. Wikimedia Foundation. en.wikipedia.org
5. Green, T. M. (2015). The Greek and Latin roots of English (5th ed.). Rowman and Littlefield.