All About Ancient Greek Accents

Paul Jungwirth February 6, 2007

Basic Rules
Accents in Ancient Greek seem tough to learn, and many teachers just skip over them. But with a little eort up front, you can learn the rules, and then as you see them applied to every word you read, they will quickly become second nature. Just think through the rules a few times and watch how they work out when applied to verbs and nouns. You might want to decline a few nouns yourself and write down the full conjugation of a verb or two. You should be comfortable with the basic rules before you study enclitics. Then get comfortable with enclitics before you go on to the strange exceptions. There are three types of accent: acute (), grave (), and circumex ().1 But they are easier to learn if we forget about the grave accent at rst. A grave accent is just an acute accent that has been ipped around by the words place in a sentence. When a word appears alone, it never has a grave accent. So rst well learn how to accents words that stand alone, considering only acute and circumex, and then well learn when you should ip acutes into graves due to sentence position. Accent position is controlled by which syllables are long or short. The meaning of long and short is not as in English. A long alpha is pronounced just like a short alpha. But the ancient Greeks actually took longer
Many examples are taken from Hanson & Quinn. In Greek, these are called oxia, varia, and perispomeni. But in modern Greek, what looks like oxia is actually called tonos (literally accent), because it represents all three.
1

to say a long alpha; they stretched it out.2 Some vowels can be either long or short, and others are always one or the other:3 short long
a a  e h i  i o u u  w

diphthongs (almost always)

The letters that are always long or always short form pairs. An eta is basically a long epsilon, and an omega is basically a long omicron (o-mega and o-micron). If an alpha has a iota subscript (), you know its long. All diphthongs are long, except for some limited circumstances which Ill note. Accents behave in two dierent ways, depending on the type of word. They can be recessive or persistent. Accents are recessive on verbs and persistent on everything else (most important, nouns and adjectives). Recessive means that the accent tries to get as far left as possible; it comes as early in the word as it can. Persistent means that the accent has a place it likes, and it wants to stay there. The nouns yuq (soul) and nsoc (island) are both persistent. In yuq, the accent wants to stay on the second syllable, but in nsoc, the accent wants to stay on the rst. This means that with nouns you must memorize the accents preferred position, but not with verbs.
 ?

Recessive Verbs

Persistent Nounish Things Nouns Pronouns Articles Adjectives Participles

So recessive and persistent tell you want an accent wants, but accents dont always get what they want. There are rules that limit where you can put accents. Whenever you write an accent, you should rst consider what the accent wants, and then apply the rules to see where its allowed. The accent may need to compromise. If necessary, it will change from acute to
Erasmian pronunciation, used by modern classics scholars, does change the sound of a vowel to indicate whether it is long or short, but this is an articial style not used by Greeks. 3 No one actually writes macrons above long vowels, but we write them here to teach which vowels are long.
2

circumex (or vice versa) to keep its spot. If that wont help, then the accent will grudgingly move to another syllable.

Possibilities
Accents can only appear on the last three syllables of a word. They never appear sooner than that. Therefore, we give special names to these three syllables. The very last syllable is called the ultima. The second-to-last syllable is called the penult. The third-to-last syllable is called the antepenult. So we can represent a word like this: -a-p-u There the a stands for antepenult, p for penult, and u for ultima. If we write a macron, it indicates a long vowel, like -a--u. If we write a breve, p it indicates a short vowel: -a--u. If we write neither, it indicates that the p vowel could be long or short. There is a rule for each syllable, telling you when an accent is allowed. First lets consider the antepenult. This is the rst place to look when you have a verb, because verbs are recessive. The antepenult can never take a circumex, and it can only take an acute when the ultima is short. If the ultima is long, the accent has to go somewhere else. So we could write that rule like this: --p- a u As long as the ultima is short, the accent can stay, but if its long, the accent has to move to the penult. Now lets talk about penults. The penult can always take an accent, but it has a rule about which accent is allowed. If the words pattern is -a-- pu (long-short), the accent must be circumex. If the pattern is anything else, the accent must be acute. We could write that rule like this: -a--u (except for -a--) p pu - -a- u p So if an accent wants to fall on the penult, consider the last two syllables. Are they long-short? Then the accent is circumex. Otherwise it is acute. Finally, lets consider the ultima. An accent can always fall here if it wants to. Note however, that verbs will never accent the ultima, because 3

being recessive, they will always want to go farther back. Only nouns accent the ultima. The rules for accent type are not very restrictive. An accent can be acute no matter what. It can be circumex only if the ultima is long: -a-p- u -a-p- u So those are the rules for where accents are allowed. Whenever you place an accent, you should rst decide which syllable it wants. Think about the rules for that syllable to see if the accent can stay there. Does it need to change its type? Must it move? antepenult n--p- a u never -a-p-u penult ultima jrwpoc -a--u (except for -a--) -a-p- p pu u -a-- pu -a-p- u

We could rephrase these rules in terms of accent type. You only need to think about this if its helpful to you. Its just another way of approaching accents: First lets talk about acute accents. An acute accent can fall on any of the three syllables: ultima, penult, or antepenult. But it can only reach the antepenult if the ultima is short: --p-. If the ultima is long, it will force a u the accent back to the penult: -a--. Acutes are less restricted when it pu comes to the penult and ultima. They can fall on the penult as long as the combination is not --. If the word is --, an accent on the penult must pu pu be circumex. Finally, acutes can always fall on the ultima, regardless of syllable lengths. Circumex accents are more restricted. They can never fall on the antepenult, and they can only fall on the penult if the word is --. They can pu fall on the ultima as long as it is long. So one consequence of these rules is that circumexes only appear on long vowels. We could diagram these basic rules like this: --p- a u -a--u (except for -a--) p pu -a-p- u - -a- u p -a-p- u 4

These rules are the heart of Greek accentuation. But to really learn them, we must put them into practice, rst on verbs and then on nouns.

Verbs
Verbs have recessive accent. Take the verb paidew. Being recessive, the accent wants to go all the way back to the antepenult if it can. But the ultima is long, so this is forbidden. Instead, it goes to the penult. But if we change the tense to aorist, we get padeusa. The ultima is short, so the accent recedes all the way back to the antepenult. In the case of verbs, these are really the only two possibilities. You never see an accent on the ultima, because the accent can always go farther back. And you never see a circumex (except for very short verbs: see below). You might think a circumex could appear on the penult, but that would require a long ultima. And if the ultima were long, the accent would recede all the way to the antepenult. Look at the ve rules and convince yourself that this is true. The only exceptions are very short verbs. For instance, jw (sacrice) has an imperative form of je. Since there is no antepenult, the accent can only go back to the penult, and it becomes a circumex due to the -- pu pattern. Note that this wont happen in indicative forms like the imperfect, because then the accent can go all the way back to the past augment: jue. So except for these very short cases, to accent a verb, just look at the last syllable. If its long, accent the penult. If its short, accent the antepenult. There are special rules for accenting contracted verbs. Position the accent as if the syllables were not contracting. This gives tim-eic and poi-ousin. Once you know the position, the accent is circumex unless the rules prohibit it. This gives dl and dhlomen, but dhlomhn and dhlomeja. These rules explain some forms of the -mi verbs, like the subjunctive did. They also help to recall the variant optative endings like lujemen.

Nouns
All other words have persistent accents. This matters most with nounish words like nouns, adjectives, articles, and pronouns, whose declensions tend to move their accents around. Suppose your word wants an ultima acute accent, like yuq. This is the simplest case, because an acute accent can appear on the ultima in all circumstances. The only complication is when 5

the accent is naturally circumex. Such words are rare, but they do exist, like g. This accent will remain a cirumex whenever possible, giving gc and gn. But suppose your word wants a penult accent. If the penult is short, as in lgoc, again everything is simple, because --u can always take a penult p acute. So even when the ultima is long, you still have lgou. If the penult is long, as in dron, the accent will change as the ending changes. This gives you drou. If the word wants an antepenult accent, it behaves much like a verb: its position is controlled by the ultima. You have njrwpoc but njrpou. Nounish words are complicated by some special rules. First, the endings -oi and -ai in third person plural forms count as short. That gives qrai and njrwpoi. Note that the dative endings -oic and -aic are not short. Neither does this rule apply to verbs, which use the -oi and -ai endings in the optative. Therefore you still have paideoi and paidesai. Second, nounish words have special rules for the genitive and dative forms (both singular and plural). Whenever these forms are accented on the last syllable, the accent becomes a circumex. Therefore delfc becomes delfo and yuq becomes yuq. This rule combines with another rule, limited to rst-declension nouns only. These words plural genitive is always accented on the ultima, no matter what. Hence, qra becomes qwrn. This last rule does not apply to feminine adjectives, nor to third-declension nouns. It is the only dierence in accenting nouns versus adjectives. These rules are sucient to explain the following declensions: Nom. S. Gen. Dat. Acc. Nom. P. Gen. Dat. Acc.
lgoc lgou lg lgon lgoi lgwn lgoic lgouc njrwpoc njrpou njrp njrwpon njrwpoi njrpwn njrpoic njrpouc delfc delfo delf delfn delfo delfn delfoc delfoc nsoc nsou ns nson nsoi nswn nsoic nsouc gor gorc gor gorn gora gorn gorac gorc qra qrac qr qran qrai qwrn qraic qrac

This table reveals another notable thing: First-declension nouns are sometimes ambiguous between the singular genitive and plural accusative, as in qrac. But when the accent falls on the ultima, they are not ambiguous, as in gorc versus gorc. 6

Articles follow the same rules, except that the masculine and feminine nominatives have no accent at all: Nom. S. Gen. Dat. Acc. Nom. P. Gen. Dat. Acc.
to t tn o tn toc toc tc t tn a tn tac tc t to t t t tn toc t

Third-declension nouns dont share all these special rules. They obey the ve regular accent rules, plus one extra: when the nominative form is a single syllable, as in ax (goat), the genitive and dative forms (singular and plural) are accented on the ultima. When this happens (and not otherwise), the genitive plural takes a circumex. That yields words like this: Nom. S. Gen. Dat. Acc. Nom. P. Gen. Dat. Acc.
ax agc ag aga agec agn ax(n) agac lpc lpdoc lpdi lpda lpdec lpdwn lpsi(n) lpdac qric qritoc qriti qrin qritec qartwn qrisi(n) qritac sma smatoc smati sma smata swmtwn smasi(n) smata

Note that participles have persistent accent, not recessive, because they are adjectives, not verbs. But usually the accent wants to fall on the stem, so this is not noticable. It only matters for a few forms:

second aorist active

-n -ntoc -ec -ntoc -c -toc -mnoc

-osa -oshc -esa -eshc -ua -uac -mnh

-n -ntoc -n -ntoc -c -toc -mnon

aorist passive

perfect active

perfect middle/passive

Grave Accents and Enclitics

When an acute accent appears on the ultima and is not followed by a pause, it becomes a grave accent. Therefore you would write, >En rq pohsen jec tn orann ka tn gn. Standard practice is to keep acutes before colons (;) and periods (.), although Ive seen one or two texts that keep acutes even before commas. So the accents would behave like so: Lgei at gun; krie, jewr ti profthc e s. And thats all you need to know about grave accents! But certain words, called enclitics, make everything more complicated. An enclitic is a word that is pronounced as almost part of the preceding word, like fle mou. Enclitics include: the indenite pronoun/adjective the personal pronouns the indenite adverbs the particles
tic, ti.

mou, moi, me, sou, soi, se. pojn, poi, pot, pou, pwc.

ge, -per, toi. te. em

the conjunction

the two-syllable present indicative actives forms of

and

fhm.

These should not be confused with proclitics, which are unaccented words pronounced closely with the next word, like or o. Because enclitics are pronounced as part of the proceeding word, they can change the accent on that word. First, they change grave accents back to 8

acute. Second, if the accent is as far forward as possible for that accent type, the enclitic adds a second acute accent to the words ultima. For instance, dik te ka asqr and nsc tic. Otherwise, the accent is unchanged. If an enclitic follows a proclitic, the proclitic gets an acute accent: E ti klyeiac, gnai, blabehc n. If enclitics follow each other, each gets an accent on its ultima except the last: Saol Saol, t me dikeic? Usually enclitics do not accent themselves. But two-syllable enclitics take an accent when they follow a verb in the -a--u pattern. This is to avoid p three unaccented syllables in a row. The accent goes on the ultima. It is acute (or grave) on short vowels and circumex on long. Hence: mhtrsi tisn and mhtrwn tinn. Enclitics following elided words also take accents: tot> st kakn. And two-syllable enclitics that begin a sentence take an accent in the same way: tinc mn lgousi, tinc d koousin. But when sti(n) begins a sentence with the meaning of there is or it is possible, it takes an accent on the penult: sti softatc tic ntaja or ok stin pofugen.

Strange Stu
Adverbs accented on the ultima always have a circumex: kakc but xwc. So if you see a word ending in -c, its probably a participle. Two-syllable prepositions sometimes switch their accent to the rst syllable when they follow their object. In poetry, p, p, met, par, per, and p, can do this. In prose, only per can. Thus, totwn pri lgwmen. Verbs have many special rules. One rule applies to verbs with a prex (pro-, po-, etc.). Often, when a verb has a prex, the accent never recedes back onto the prex. This applies to forms with a past indicative augment, hence the aorist pljon but the imperative pelje. It applies to the perfect, hence sunqa. And it applies to innitives and participles, hence katalsai, katalon and katajec. Those are the important cases to remember. There are a variety of other cases that are more unusual and not necessary to memorize. Many innitives are weird. Here are the rules:

present active present middle/passive future active future middle future passive rst aorist active rst aorist middle second aorist active second aorist middle aorist passive perfect active perfect middle/passive

lein lesjai lsein lsesjai lujsesjai lsai lsasjai balen balsjai lujnai leluknai lelsjai

Always acute on the antepenult. Always Always Always Always Always Always Always Always Always acute on the antepenult. acute on the antepenult. circumex or acute on the penult. acute on the antepenult. circumex on the ultima. acute on the penult. circumex on the penult. acute on the penult. acute on the penult.

Two funny verbs are em and fhm. In the present indicative, the twosyllable forms are not recessive but enclitic. This gives sentences like edamonc ste. em and fhm also take a circumex in all forms of the subjunctive: , c, , etc. and f, fc, f, etc.

10