Accent and morae

In every Greek word one syllable stands out from the others. Every language with a word accent does something like this, but languages differ in how they make the syllable stand out. This page explains how ancient Greek did it, and then how the accent marks in our texts record it.

Stress and pitch

English makes a syllable stand out mainly by stress. A stressed syllable is said with more effort: it is louder and longer, as the first syllable of photograph is, or the second of photographer. English stressed syllables often carry a change of pitch as well. That pitch is not part of the word, though. It changes with what the speaker means: the same word can be said with a fall or with a rise, as a statement or as a question.^[1]

A pitch accent works differently. The accented syllable is said on a higher note than the others. A school grammar of 1902 put the contrast in its simplest form. A language with a pure stress accent would say every syllable on the same note but the accented one louder. A language with a pure pitch accent would say every syllable equally loud but the accented one higher.^[2]

The ancient Greek sources describe their accent in terms of pitch. High pitch is the one feature of the accent they notice.^[3] The accented syllable was ὀξύς, “sharp”, which in Greek also means high in pitch. The other syllables were βαρύς, “heavy”, which means low. The word βαρύς cannot mean “quiet”, because it is used of a sound that is both low and loud, as in βαρυβρεμέτης, “deep-thundering”, an epithet of Zeus.^[4] The words for the accent itself, τόνος and τάσις, “stretching”, come from the tuning of a string.^[4] The general word was προσῳδία. It means a song sung to the words, and the Latin accentus is a translation of it.^[5]

Here is ἄνθρωπος with two kinds of prominence. In the first clip, Yokocentaur reads it in reconstructed Classical Attic: the first syllable is higher in pitch than the following syllables. The second clip is a synthetic demonstration of an English-like stress on that syllable: louder and longer, on one note. Compare the pitch movement; the two clips have different voices.

How to read the diagrams

Every diagram on this page has three rows. At the top are the syllables of the word in Greek. At the bottom is one box for each mora. A mora is a unit of vowel length: a short vowel has one, a long vowel or a diphthong has two. The Morae tab explains them. The box on which the voice is high is filled red. Over each box a small stroke shows the pitch: a rising stroke on the high mora, a falling stroke on the mora after it, and a level stroke on the others.

The blue line in the middle is the pitch of the clip, measured with Praat. Praat is a free program that phoneticians use to measure speech. Where the line breaks, the voice was silent or the sound was a consonant without voice, such as π or τ. Press the round button under a diagram to hear the clip. A dot then runs along the line, and the box being spoken is outlined.

One high syllable in each word

The critic Dionysius of Halicarnassus wrote about the melody of speech. He says that a Greek word has only one high syllable, and that the others are low.^[6] He also gives the size of the step. The voice does not rise more than about a fifth, the interval of three tones and a semitone.^[7]

The voice does not jump up to the high syllable and down again. It climbs towards the high gradually and falls back after it.^[8] The musical settings that follow the accent show the shape in more detail. Syllables before the accent rise steadily towards it. The fall after the peak is larger than the rise. Most of the fall comes at once, on the first mora after the peak; later syllables sink more gently to the end of the word.^[9]

Accents in a sentence

A word’s high was not always the same height. In a sentence, each accent had to fit into the melody of the whole phrase. A phrase here means a clause, or a long participle phrase, said as one unit. The two hymns inscribed at Delphi show how this worked, because their melodies follow the accents closely.

In the Delphic hymns the first full accent of a phrase is usually its highest note. Where the setting survives, the first acute or circumflex is the highest pitch of its phrase 71 times in 100.^[10] After it, each accent tends to be set a little lower than the one before. On average the step down from one word’s high to the next is just over a semitone.^[11] This gradual lowering through a phrase is called downtrend. Most modern languages show it too, whether they use stress, a pitch accent or tones.^[12]

The lowering stops at the end of the phrase. At the start of a new phrase the melody is set higher again. In the hymns this reset is a rise of just over a whole tone, from the last accent of one phrase to the first accent of the next.^[13] Smaller word groups inside a phrase get a smaller lift.^[14]

Two things interrupt the lowering. A word with a grave does not push the next accent down.^[15] The names of gods and places are set higher than the word before them, as if spoken with emphasis.^[16]

This picture rests on two inscribed hymns, and it is Devine and Stephens’ reconstruction. Later songs show little of it.^[17]

About the sounds

The human readings of ἄνθρωπος and ἀνθρώπου are modern reconstructions, not surviving ancient voices. Their pitch lines are measured from the recordings. Each human trace is fitted to its box so its full rise and fall can be seen; compare the shape, rather than the height of different voices. These modern readings illustrate accent position, not an exact ancient pitch range. The other clips are synthetic, marked ∿: a formant synthesiser speaks the sounds in the reconstructed values of classical Attic.^[18] The vowel lengths and the pitch are then set with Praat. A long vowel lasts twice as long as a short one. The loudness of the vowels is evened out, so that no syllable is stressed. The pitch follows the shape just described: a rise to the high mora, a steep fall on the next mora, and a gentle fall to the end of the word. The whole movement stays within a fifth. The clips show the rules in sound; they are not evidence for how a Greek actually spoke.

Notes

1. ^ Probert 2006, p. 56.
2. ^ Thompson 1902, p. 427.
3. ^ Probert 2006, p. 55.
4. ^ ^{a b} Allen 1968, p. 106.
5. ^ Smyth 1920, p. 37.
6. ^ Devine and Stephens 1994, p. 172.
7. ^ Allen 1968, p. 109; Devine and Stephens 1994, pp. 171–172.
8. ^ Allen 1968, p. 113.
9. ^ Devine and Stephens 1994, pp. 183–189.
10. ^ Devine and Stephens 1994, p. 441.
11. ^ Devine and Stephens 1994, p. 442.
12. ^ Devine and Stephens 1994, pp. 435–436.
13. ^ Devine and Stephens 1994, p. 450.
14. ^ Devine and Stephens 1994, p. 408.
15. ^ See the Three marks tab.
16. ^ Devine and Stephens 1994, pp. 479–480.
17. ^ Devine and Stephens 1994, pp. 451–452.
18. ^ See Sounds.

No ancient Greek speaker can be heard today, and the classical Greeks did not write accents. Yet we know the accent of almost every word. This tab explains where that knowledge comes from.

No marks in classical times

Greek inscriptions of classical times, the fifth and fourth centuries BC, show no accents.^[1] Speakers knew the accent of every word, because it was part of the way they spoke, so they had no need to write it.^[1] Writers did sometimes talk about the accent, as the Accent and meaning tab shows.

The grammarians

The marks were devised at Alexandria about 200 BC. Their invention is credited to Aristophanes of Byzantium, the librarian there.^[2] An ancient note says that the marks were made for two purposes. One was to tell apart words that could be misread. The other was to show the melody of the voice, as if speech were being sung.^[3]

The scholars who followed him discussed the accent of many words, especially in Homer. Aristarchus, the next librarian, who died about 144 BC, is the one most often quoted.^[4] In the second century AD the grammarian Herodian gathered the teaching into one very large work. All later grammarians depend on him, directly or indirectly.^[5] Most of what we know comes from this tradition, and from the accents in medieval manuscripts that follow it.^[6]

Papyri

Some papyri of literary texts carry accent marks from the second century BC onwards.^[7] Most mark only a few words, often those that could be misread. They use the same three signs, though not always in the modern way. One system put a grave on every syllable before the accent, so that νὲκρον means νεκρόν.^[8]

Where a papyrus marks an accent, it almost always stands where the grammarians and the medieval manuscripts put it. In one sample of Homeric papyri, 131 accented words of one type were checked, and only one accent was misplaced.^[9] So the system is at least as old as the second century BC.^[8]

Manuscripts

Accents were rarely written in manuscripts in capital letters. From the ninth century AD, when scribes began to write in small letters, every word got its accent.^[10] The scribes did not simply write the accents of their own speech. Notes in the margins name the grammarians they used, Herodian among them.^[11]

Music

Some songs survive with their melody written above the words, on stone or papyrus. Many of them set the accented syllable on the highest note of its word, as the three marks tab shows. The earliest song that follows the accents in this way is of the third century BC.^[12] Songs in stanzas generally do not follow them, because the same tune served stanzas with different words.^[12]

Vedic Sanskrit

Vedic Sanskrit, the language of the ancient Indian hymns, also had a pitch accent, and its accent often falls on the same syllable as the Greek.^[13] Greek πατήρ, “father”, has the accent on its last syllable, and so has Vedic pitā́. The vocatives πάτερ and pítar both draw it back to the first.^[13] Such matches show that many Greek accents had stayed in place since long before Homer.^[14] The rules that limit the accent are also older than the Greek dialects.^[15]

So the written accents are a faithful record of the word melodies of ancient Greek.^[16] Whether the accent also had some loudness in it is debated.^[17]

From pitch to stress

The pitch accent did not last. Greek lost its distinction between long and short vowels, and the accent became a stress.^[18] The change cannot be dated exactly. In everyday speech it was widespread by the middle of the second century BC.^[19] It was complete by the fourth century AD.^[20]

The stress stayed on the syllable that had carried the high pitch. So Modern Greek still stresses most words where the ancient mark stands.^[21] In Modern Greek all three marks came to mean the same thing, a stress.^[22]

Notes

1. ^ ^{a b} Allen 1968, p. 114.
2. ^ Probert 2006, p. 21; Allen 1968, p. 114.
3. ^ Probert 2006, p. 21.
4. ^ Probert 2006, pp. 22–23.
5. ^ Probert 2006, p. 22.
6. ^ Probert 2006, p. 15.
7. ^ Probert 2006, p. 45.
8. ^ ^{a b} Probert 2006, p. 46.
9. ^ Probert 2006, pp. 46–47.
10. ^ Probert 2006, pp. 48–49.
11. ^ Probert 2006, p. 49.
12. ^ ^{a b} Probert 2006, p. 48.
13. ^ ^{a b} Allen 1968, p. 107.
14. ^ Probert 2006, pp. 83–84.
15. ^ Probert 2006, pp. 84–86.
16. ^ Allen 1968, p. 118.
17. ^ See Accent: pitch or stress?
18. ^ Horrocks 2010, p. 169; Probert 2006, pp. 50–51.
19. ^ Horrocks 2010, p. 169.
20. ^ Allen 1968, p. 119.
21. ^ Probert 2006, p. 86.
22. ^ Horrocks 2010, p. xx.

The Greek accent rules count something smaller than a syllable. They count morae. Once you can count morae, every rule on the next tabs becomes simple arithmetic.

What a mora is

A mora is a unit of length in a vowel. A short vowel has one mora. A long vowel has two. A diphthong has two as well.^[1] A diphthong is two vowel sounds said in one syllable, such as αι or ευ.^[2]

The high pitch of the accent sits on one mora, not on a whole syllable.^[3] A short vowel has only one mora, so the high can only be on it. A long vowel has two morae, so the high can be on the first or on the second. That is why Greek needs two marks for an accented long vowel: the acute and the circumflex, explained on the next tab. A language that can accent a long vowel in two ways like this is a typical pitch-accent language.^[4]

Which vowels are long

Two vowels are always short: ε and ο. Two are always long: η and ω.^[5] The other three, α, ι and υ, are short in some words and long in others.^[5] These three are called the δίχρονα, “vowels of two lengths”.^[6] The spelling does not show their length. Grammars and dictionaries mark a long one with a macron, as ᾱ, and this page does the same where it matters.

Every diphthong counts as long. So do ει and ου, which in classical Attic were single long vowels written with two letters, and the vowels with iota subscript, ᾳ ῃ ῳ.^[7]

The accent itself sometimes shows the length of α, ι or υ. A circumflex stands only on a long vowel, so the α of πᾶς, “all”, must be long.^[8] The rules on the next tabs give other clues of the same kind.^[9]

Consonants do not count

Pitch is carried by the voice. Vowels are voiced, and so are sounds like λ, μ, ν and ρ, but a voiceless consonant such as π, τ, κ or σ has no pitch at all.^[10] So the accent rules count only the morae of vowels. A consonant after a short vowel adds nothing.

This is the main difference between the morae of the accent and the length of syllables in metre. In metre a syllable is long if it has a long vowel or if it is closed by a consonant.^[11] Metricians call a short syllable one mora and a long one two.^[12] The accent ignores the closing consonant.^[13] Take φοῖνιξ, “palm tree”. Its last syllable, -νιξ, is long in verse, because it ends in two consonants (ξ is κ + σ). For the accent it is short, because its vowel is a short ι.^[14] The circumflex on φοῖ- shows this, as the rules will explain.

The diagram sets the two counts side by side for five syllables. The top row of boxes is the metre’s count and the bottom row the accent’s. Arrows run from each vowel to its boxes, and from a closing consonant to the box it adds for the metre only. Over the accent’s boxes a small stroke shows the pitch: it rises on the high mora, rises and falls over the two morae of a circumflex, and stays level on the others.

a mora of the vowela mora made by a closing consonant (metre only)ε vowelπ closing consonant

Over the accent’s boxes, the pitch:high: the voice rises on this moracircumflex: up and down on one long vowellevel, low

The rule in one line: for the accent, count only the vowel: a short vowel is one mora, a long vowel or a diphthong two. The words below show it with their pitch.

Final -αι and -οι

There is one large exception to the counting. At the very end of a word, the diphthongs -αι and -οι usually count as one mora for the accent.^[15] So ἄνθρωποι, “men”, keeps its accent on the first syllable, as if it ended in a short vowel. The diagrams show such a final mora as a dashed box.

They count as two morae in three cases. The first is the optative, a verb form, as in παιδεύσαι, “may he educate”. The second is old locative forms, such as οἴκοι, “at home”. The third is some interjections, such as αἰαῖ, “alas”.^[16] In metre, final -αι and -οι are long like any other diphthong.^[16]

Count them yourself

Choose a word. For each syllable, decide whether its vowel has one mora or two, then press Check. The answer explains each syllable and shows where the rules let an accent stand on a word of that shape.

Notes

1. ^ Probert 2006, pp. 57–58; Allen 1968, p. 81.
2. ^ See Diphthongs.
3. ^ Probert 2006, p. 58.
4. ^ Probert 2006, pp. 55–56.
5. ^ ^{a b} Smyth 1920, p. 8.
6. ^ See Vowels.
7. ^ Allen 1968, p. 81.
8. ^ Smyth 1920, p. 36.
9. ^ Smyth 1920, p. 40.
10. ^ Allen 1973, pp. 83–84.
11. ^ See Syllables and quantity.
12. ^ West 1987, p. 7.
13. ^ Allen 1973, p. 204.
14. ^ Allen 1968, pp. 106–107.
15. ^ Probert 2006, p. 61; Smyth 1920, p. 39.
16. ^ ^{a b} Probert 2006, p. 61.

Greek texts carry three accent marks. Each tells you where the voice was high in a word, and how the high was placed on the morae of its vowel.

Three signs

The three marks are the acute (ά), the circumflex (ᾶ) and the grave (ὰ).^[1] The Greek names describe pitch. The acute is the ὀξεῖα, the “sharp” or high one. The circumflex is the περισπωμένη, the one “bent round”, because it goes up and comes down.^[2] The grave is the βαρεῖα, the “heavy” or low one.^[3]

A mark stands over the vowel of its syllable. In a diphthong it stands over the second letter, as in τοῦτο and αὐτός. A breathing comes before an acute or grave and under a circumflex, as in οἵ and οὗτος.^[1]

The marks are not as old as the language. Their invention is credited to Aristophanes of Byzantium, the librarian at Alexandria, about 200 BC.^[4] In the early papyri they appear only here and there, mostly where a word could be misread. One early system put a grave on every low syllable, as in Θὲόδὼρὸς.^[5] Manuscripts mark every word only from the ninth century AD.^[6]

The acute

On a short vowel the acute means that the voice is high on that vowel. The fall comes on the next syllable.^[7] In λόγος, “word”, the voice is high on λό- and falls on -γος.

On a long vowel or a diphthong the acute means that the high is on the second mora. The voice rises through the vowel and reaches its high at the end.^[8] In ἀγαθή, “good” (feminine), the η rises.

The circumflex

The circumflex stands only on a long vowel or a diphthong.^[1] It means that the high is on the first mora, and that the voice falls on the second.^[9] The rise and fall happen inside one syllable. The grammars describe the circumflex as an acute and a grave joined together.^[10]

Compare ἀγαθῶν, “of good things”, with ἀγαθή. Both have a long final vowel with the high on it. In ἀγαθή it is on the second mora and the vowel rises. In ἀγαθῶν it is on the first mora and the vowel rises and falls. Two forms of one verb differ only in this way: λῆξαι, “to cease”, and λήξαι, “may he cease”.^[11]

The grave

The grave is not a third kind of accent. It is the acute of a word’s last syllable, written differently when another word follows in the same sentence.^[12] Διός, “of Zeus”, has an acute when it stands alone, but in Διὸς υἱός, “son of Zeus”, it has a grave.^[13] The acute stays before a punctuation mark, before an enclitic,^[14] and on the question word τίς, “who?”.^[15]

What the grave meant in speech is disputed. In the songs whose music survives, a syllable with a grave is not set lower than the rest of its word. It rises about a third of a tone, where an acute rises nearly a whole tone.^[16] On this evidence the grave is likely to have been a lowered high. It was higher than the unaccented syllables before it, but lower than a full accent.^[13] One explanation is that a Greek word kept the fall after its high inside its own bounds. A final high had no room for its fall, and so it was lowered.^[17] An older view, Laum’s, held that the grave was only a habit of writing, but objections to it were raised at once.^[13]

The Byzantine grammarian George Choeroboscus described the change in a picture. In connected speech, he says, every oxytone “puts its acute to sleep” and turns it into a grave, except the question word τίς.^[18] The music shows the exception. In one late song τίς is set a whole tone higher than the accent of the next word, νόστος, “return”.^[19]

In the Delphic hymns a grave also changes the melody of the phrase. After a full accent, the next accent in a phrase is normally set a little lower. When a word with a grave comes between them, the next accent is set higher instead, by about a semitone on average.^[20] The grave word and the word after it seem to form one rising line towards the next full accent.^[21] The first tab explains this fall through the phrase.

What the music shows

Some songs survive with their melody written above the words. In those composed to follow the accent, three habits stand out. The accented syllable gets a note at least as high as any other in its word. A circumflex syllable is often sung on two notes, the first higher. After a final acute that is not at a pause, the melody does not fall again until the next accent.^[22] The two hymns inscribed at Delphi follow these habits strictly.^[23] An acute on a long vowel is sometimes sung on a rising pair of notes, the mirror of the circumflex.^[24]

The epitaph of Seikilos, a short song inscribed on a tombstone, shows the same thing. Its acute syllables, in ὅλως, ὀλίγον and χρόνος, are sung higher than the rest of their words.^[25] Its circumflexes, in λυποῦ, ζῆν and ἀπαιτεῖ, are sung on two falling notes.^[26] You can hear its words one by one on the Practice tab, and the song itself in the Music section.^[27]

Names for words by their accent

Grammars name a word by the mark it carries and where the mark stands.^[28] The last syllable is the ultima, the one before it the penult, and the one before that the antepenult.

A word with no accent on its last syllable is also called barytone.^[10]

Notes

1. ^ ^{a b c} Smyth 1920, p. 37.
2. ^ Smyth 1920, p. 38; Allen 1968, p. 115.
3. ^ Allen 1968, pp. 106, 115.
4. ^ Probert 2006, p. 21; Smyth 1920, p. 38.
5. ^ Allen 1968, p. 115; Probert 2006, pp. 45–46.
6. ^ Probert 2006, pp. 48–49.
7. ^ Probert 2006, p. 58; Horrocks 2010, p. xx.
8. ^ Probert 2006, p. 58; Allen 1968, p. 115.
9. ^ Probert 2006, p. 58.
10. ^ ^{a b} Smyth 1920, p. 38.
11. ^ Allen 1968, p. 115.
12. ^ Probert 2006, pp. 58–59; Smyth 1920, p. 37.
13. ^ ^{a b c} Probert 2006, p. 59.
14. ^ See the Enclitics tab.
15. ^ Smyth 1920, p. 37; Probert 2006, p. 59.
16. ^ Devine and Stephens 1994, pp. 181–182.
17. ^ Devine and Stephens 1994, pp. 180–181; Allen 1968, pp. 115–116.
18. ^ Devine and Stephens 1994, pp. 454–455.
19. ^ Devine and Stephens 1994, p. 455.
20. ^ Devine and Stephens 1994, pp. 445–446.
21. ^ Devine and Stephens 1994, pp. 448–449.
22. ^ Probert 2006, p. 48.
23. ^ Devine and Stephens 1994, p. 173.
24. ^ Devine and Stephens 1994, p. 193.
25. ^ Allen 1968, p. 109.
26. ^ Allen 1968, p. 112.
27. ^ See Melody.
28. ^ Smyth 1920, p. 38; Probert 2006, p. 62.

The Greek accent is free. Nothing in the sound of a word tells you which syllable carries it, and the place has to be learned with the word.^[1] But it is free only within limits. A small set of rules says where an accent may stand, and which mark it must take there.^[2] The rules look only at the end of the word, and they count morae.^[3]

The last three syllables

The accent can stand only on one of the last three syllables of a word.^[4] These three have Latin names. The last syllable is the ultima. The one before it is the penult, “almost last”. The one before that is the antepenult, “before the penult”.

The law of limitation

The rules that limit the accent are together called the law of limitation. In plain words there are four of them.^[5]

1. An acute can stand on the antepenult, the penult or the ultima, but no further back.
2. A circumflex can stand on the penult or the ultima, but never on the antepenult.
3. If the vowel of the ultima is long, no accent can stand on the antepenult. The same holds if the ultima ends in two consonants, as in -ξ or -ψ.
4. If the vowel of the ultima is long, a circumflex cannot stand on the penult.

So εὐχόμενος (euchomenos), “praying”, is a possible word, but *εὔχομενος is not, because the acute would stand four syllables from the end. ἄνθρωπος, “human being”, has an acute on the antepenult, which is allowed while the last vowel is short. In the genitive ἀνθρώπου the last vowel is long, so the accent must move forward to the penult.^[2]

The σωτῆρα rule

One more rule decides between acute and circumflex on the penult. If the penult has a long vowel and carries the accent, and the vowel of the ultima is short, the mark must be a circumflex.^[6] So the accusative of σωτήρ, “saviour”, is σωτῆρα, never *σωτήρα. The rule is named after this example. It is also called the “final trochee rule”, but that name is misleading, because the rule is about the length of vowels, not of syllables. νύκτα, “night”, ends like a trochee, long then short, but its penult has a short vowel and keeps its acute; κλῖμαξ, “ladder”, has a closed last syllable and still takes the circumflex.^[7]

When the ultima is long, the same accented penult takes the acute, by rule 4. So the nominative δῶρον, “gift”, becomes δώρου in the genitive.

One picture for all the rules

The rules can be seen as a single limit on the fall that follows the high. Think of the high mora together with the fall after it as one unit. On a circumflex this unit is one syllable. After an acute it runs on through the next syllable. After this unit, the word may go on for at most one more mora.^[8] Count the morae after it and the rules follow. In ἄνθρωπος only the short -ος follows the fall on -θρω-, so it is allowed. In *ἄνθρωπου two morae would follow, so the accent moves to ἀνθρώπου. The σωτῆρα rule is a separate rule even in this picture.^[9]

In the diagrams below, the numbers under the boxes count the morae from the end of the word. A dot marks a final -αι or -οι mora, which the accent does not count.

Recessive and persistent

Within these limits, two habits decide the place of the accent in most words.

A recessive accent goes as far back from the end as the rules allow.^[10] Finite verbs are recessive: once you know the length of the last vowel, you know where the accent goes.^[11] In παιδεύω, “I educate”, the last vowel is long, so the accent goes no further back than the penult. In παιδεύομεν, “we educate”, the last vowel is short, so it goes back to the antepenult.

A persistent accent stays on the syllable where it stands in the dictionary form, and moves only when the rules force it.^[12] Nouns and adjectives behave this way. The name “persistent” comes from teaching grammars.^[13] The Nouns and adjectives and Verbs tabs show both in full paradigms.

Exceptions

A few Attic and Ionic words break the law of limitation. πόλεως, “of a city”, has an acute on the antepenult before a long final vowel.^[14] The word was once πόληος, which obeyed the law. Later the two vowels exchanged their lengths, and the accent stayed where it was.^[15] This shows that the law is older than the exchange, which already appears in Homer.^[16]

The genitive plural πόλεων, “of cities”, is a similar case. The ancient grammarians knew it as an Attic form. Herodian, the great authority on accents in the second century AD, judged that the regular form was πολέων, accented on the penult. He recorded πόλεων as an Attic irregularity.^[17] A century later the Atticists, who held classical Attic to be the model of good Greek, reversed the judgement. For them πόλεων was correct because it was Attic.^[18] The learned form won, and Modern Greek still says πόλεων.^[19]

The grammarians noted other Attic accents that broke their rules. Herodian’s father, Apollonius Dyscolus, observed that Attic speakers drew the accent back in χάριεν and ἄληθες, “really?”. An ancient note on the Iliad records Attic πῆρον, “maimed”, where Homer has πηρόν.^[20]

Other dialects had other habits. Lesbian, the dialect of Sappho and Alcaeus, made almost every word recessive.^[21] Doric had no σωτῆρα rule and counted final -οι as long.^[22]

Notes

1. ^ Smyth 1920, p. 38; Probert 2006, p. 60.
2. ^ ^{a b} Probert 2006, p. 60.
3. ^ Allen 1968, p. 114.
4. ^ Smyth 1920, p. 37.
5. ^ Probert 2006, p. 60; Smyth 1920, p. 39.
6. ^ Probert 2006, p. 61; Smyth 1920, p. 39.
7. ^ Probert 2006, p. 61.
8. ^ Allen 1968, p. 114; Allen 1973, p. 237.
9. ^ Allen 1973, pp. 237–238.
10. ^ Smyth 1920, p. 38; Probert 2006, p. 62.
11. ^ Probert 2006, pp. 86–87.
12. ^ Smyth 1920, p. 47.
13. ^ Probert 2006, p. 65.
14. ^ Smyth 1920, p. 39.
15. ^ Probert 2006, p. 85.
16. ^ Probert 2006, pp. 85–86.
17. ^ Probert 2011, pp. 277–279.
18. ^ Probert 2011, p. 279.
19. ^ Probert 2011, p. 278.
20. ^ Probert 2011, p. 276.
21. ^ Smyth 1920, p. 38; Probert 2006, pp. 72–73.
22. ^ Probert 2006, pp. 71–72; see the Lesbian Aeolic and Doric pages.

Some short Greek words have no accent of their own. They are said together with a neighbouring word, as if the two were one word. English does the same with a and the: in a banána [ə bəˈnɑːnə] only the noun has an accent.^[1]

Two kinds of leaning word

A proclitic leans forward onto the word after it. An enclitic leans back onto the word before it.^[2] The word an enclitic leans on is called its host.

There are only ten proclitics. They are the forms of the article that begin with a vowel, ὁ, ἡ, οἱ, αἱ; the prepositions ἐν, εἰς and ἐκ (ἐξ); εἰ, “if”; ὡς, “as”; and οὐ, “not”.^[3] A proclitic does not change the word after it.

In each example below the proclitic carries no mark and no high pitch of its own. It is said together with the next word, as if the two were one word, and the voice rises only on that word’s accented syllable.^[4] Each diagram shows the morae of the group, the high mora filled, and the pitch of the clip.

The same words occur in every page of Greek. Here are some of them in a real reading of Homer, by David Chamberlain.

The rule in one line: a proclitic has no accent of its own and is said as part of the word after it.

The enclitics are more numerous:^[5]

• the unemphatic personal pronouns μου, μοι, με, σου, σοι, σε, and οὑ, οἱ, ἑ;
• the indefinite τις, τι, “someone, something”, in all its cases, and the indefinite adverbs που, “somewhere”, ποτε, “ever”, πως, “somehow” and their like;
• the present indicative of εἰμί, “I am”, and φημί, “I say”, except the forms εἶ and φῄς;
• the particles γε, τε, τοι and περ.

Here is one of each kind in use. In every example the enclitic has no accent of its own, and the voice falls onto it from the high of the word before.^[6]

The interrogatives that look like the indefinites are not enclitic. τίς, “who?”, ποῦ, “where?”, πότε, “when?” and πῶς, “how?” always keep their accent.^[7] The music shows the difference. In εἴ τις, “if anyone”, the pitch falls from εἰ onto τις; to the interrogative τίνος, “whose?”, it rises steeply.^[8]

What an enclitic does to its host

An enclitic loses its accent and forms one unit with its host. The accent rules then apply to the host and the enclitic together, as if they were one word.^[9] That is why the host’s accent changes in some cases and not in others. There are five cases.^[10]

1. An oxytone keeps its acute. It does not change it to a grave: καλόν ἐστι, “it is fine”.
2. A perispomenon stays as it is: φιλῶ σε, “I love you”.
3. A proparoxytone or a properispomenon gains a second acute on its last syllable: ἄνθρωπός τις, “some man”; δῶρόν ἐστι, “it is a gift”.^[11] Without it, the high would stand four syllables or too many morae from the end of the combined word.
4. A paroxytone gains nothing. A one-syllable enclitic after it loses its accent: φίλος μου, “my friend”. A two-syllable enclitic keeps its accent, on its last syllable: φίλοι τινές, “some friends”. The rule avoids two acutes on neighbouring syllables, as in *φίλοί τινες.^[9]
5. A proclitic gains an acute: εἴ τις, “if anyone”; ἔν τινι, “in something”.^[12]

An enclitic’s accent, when it moves onto the host, is always written as an acute.^[7]

Why: the fall goes onto the enclitic

The rules make sense in terms of pitch. Within a word the voice falls after the high. Between full words it does not: the fall stays inside its own word. An enclitic belongs to its host, so the fall can run onto it.^[11] In ἀγαθός ἐστιν the voice is high on -θός and falls on ἐσ-. That is why the oxytone keeps its acute: it has somewhere to fall.

After a proparoxytone the combined word is too long. In ἄνθρωποί τινες more than one mora would follow the fall, which the rules forbid, so a second high is needed. In μεγάλοι τινές the second accent goes onto the enclitic, not *μεγάλοί τινες. A high straight after a high would leave μεγάλοι without its fall.^[11] The music confirms the fall onto an enclitic. After τάν the melody drops two whole tones onto τε.^[8]

When an enclitic keeps its accent

An enclitic is said with its own accent, or orthotone, when it is emphatic, as in ἢ σοὶ ἢ τῷ πατρί σου, “either to you or to your father”, and when it begins a sentence.^[9]

The verb ἔστι, “is”, is the commonest case. As an enclitic it is written ἐστι, without an accent of its own. It is written ἔστι, with an acute on its first syllable, in these cases:^[13]

• when it begins a clause;
• when it means “there is”, or “it is possible”;
• after οὐκ, μή, εἰ, ὡς, καί, ἀλλά (ἀλλ’) and τοῦτο (τοῦτ’).

There is an example of every case below. Four are quoted from poems on this site, and Smyth gives the one with εἰ himself. The rest are our own short sentences. In each, ἔσ- carries the high pitch.

The rule in one line: ἔστι keeps its accent at the start of a clause, when it means “there is” or “can be”, and after these seven words. The music agrees: in a surviving song the syllable ἔσ- of οὐκ ἔστ᾽ is set higher than the accented syllable of the next word.^[14]

Emphatic and unemphatic pronouns

The personal pronouns show the choice most clearly. Each of them has two forms. The orthotone form keeps its own accent: ἐμοῦ, ἐμοί, ἐμέ, σοῦ, σοί, σέ. The enclitic form leans on the word before it: μου, μοι, με, σου, σοι, σε. In the first person the orthotone forms also begin with ἐ-.^[15]

The choice depends on emphasis. A pronoun that sets one person against another is orthotone. The ancient grammarian Herodian gives the pair ἐμοῦ ἤκουσας, “you heard me”, that is, not someone else, and ἤκουσά σου, “I heard you”.^[15] The ancient notes on the Iliad mark σοί as orthotone in εἰ δὴ σοὶ πᾶν ἔργον ὑπείξομαι, “if I am to give way to you in everything” (Iliad 1.294), for this reason.^[15]

The surviving music agrees. In one song the emphatic κἀμέ, “me too”, is set on a rising third, higher than the accent of the next word. In the first Delphic hymn the unemphatic νιν, “him”, is set like an unaccented word ending.^[16]

A subject pronoun such as ἐγώ, “I”, or σύ, “you”, is never enclitic. Greek leaves the subject unexpressed unless it matters, so a subject pronoun that is spoken at all is emphatic.^[17]

The school grammars give the rule for prose in the same terms. The two-syllable forms ἐμοῦ, ἐμοί, ἐμέ are always emphatic in prose, and usually in verse. The one-syllable forms μου, μοι, με are usually enclitic.^[18] Even the indefinite τις can keep an accent when it is emphatic. It does so when it means “somebody” as opposed to “nobody”, as in ηὔχεις τὶς εἶναι, “you boasted that you were somebody”.^[19]

Position matters as well. At the start of a clause even an unemphatic pronoun is orthotone. The grammarian Apollonius Dyscolus compared ἐμὲ δ᾽ ἔγνω, “and he knew me” (Odyssey 11.91), where the pronoun begins the clause, with ἔγνω δὲ ψυχή με, “and the spirit knew me” (Odyssey 11.471), where it does not.^[20] An enclitic cannot lean across a break in the sentence. After a pause, therefore, the orthotone form is used; in Homer it is also common at the start of a line.^[20]

More than one enclitic, and fused words

When several enclitics come in a row, each passes an accent to the word before it, and only the last has none: εἴ πού τίς τινα ἴδοι, “if anyone ever saw anyone anywhere”.^[9] This is the rule of the ancient grammarians. Even they did not apply it every time, and most modern scholars take it to be a convention of writing rather than a description of speech.^[21]

Some enclitics grew together with the word before them into one word. These compounds are still accented as if the two parts were separate: ὥστε, “so that”, οὔτε, “and not”, εἴτε, “or if”.^[9] They look as if they break the σωτῆρα rule, but they do not.

A proclitic with an accent

A proclitic takes an accent in a few places.^[7] οὐ is accented at the end of a sentence: πῶς γὰρ οὔ; “for why not?”. It is also accented when it stands alone as “no”. In poetry ἐξ, ἐν and εἰς take an acute when they follow their noun at the end of a line, and ὡς becomes ὥς when it follows its noun: θεὸς ὥς, “like a god”. Before an enclitic a proclitic takes an acute, as in εἴ τις above.

Were proclitics really unaccented?

This question is disputed. The ancient grammarians had no class of proclitics. The word was coined by the German scholar Gottfried Hermann in 1801, for the short words that the manuscripts write without an accent mark.^[22]

The music suggests that these words were not completely toneless. In the Delphic hymns a proclitic is usually set on a level note a little below the accent of the next word. At the start of a sentence it can carry a large rise: ἀλλά, “but”, rises a whole octave in the second hymn.^[23] Devine and Stephens conclude that proclitics kept a reduced accent, like the grave of other small words, rather than none at all.^[24]

Anastrophe: a preposition after its noun

A two-syllable preposition is normally accented on its last syllable: ἀπό, παρά, περί, ὑπό. In a few positions the accent stands on the first syllable instead. This is called anastrophe, “turning back”.^[25]

The accent turns back in three cases:^[25]

1. When the verb is left out and the preposition stands for a compound verb: ἐπεὶ οὔ τοι ἔπι δέος, “since there is no fear in you” (Iliad 1.515). Here ἔπι stands for ἔπεστι, “is in”.
2. When the preverb follows its verb: ἔχεν κάτα γαῖα μέλαινα, “the black earth held him down” (Iliad 2.699).
3. When the preposition follows the noun it governs: ἄστυ κάτα, “throughout the city”, Herodian’s example.

Nothing may come between the noun and the preposition. Even a small elided word blocks the change, as in τῷ δ᾽ ἐπί (Iliad 7.163).^[26] The accent on the first syllable is the older one. It matches Sanskrit ápa, párā, pári and úpa.^[25]

Not every preposition can follow its noun. πρό, ἀντί, διά, ἀνά and ἀμφί never do, and ἐν does so only in epic.^[27] The forms πάρα, ἔπι, μέτα, πέρι, ὕπο and ἔνι, standing for πάρεστι, ἔπεστι and the like, differ from the ordinary prepositions only in their accent.^[28]

Notes

1. ^ Probert 2006, pp. 69–70; Roach 2004, p. 243.
2. ^ Probert 2006, pp. 69–70.
3. ^ Smyth 1920, p. 41.
4. ^ Probert 2006, pp. 69–70; Smyth 1920, p. 41.
5. ^ Smyth 1920, p. 42; Devine and Stephens 1994, pp. 353–355.
6. ^ Allen 1968, p. 116; Smyth 1920, p. 43.
7. ^ ^{a b c} Smyth 1920, p. 42.
8. ^ ^{a b} Devine and Stephens 1994, p. 354.
9. ^ ^{a b c d e} Smyth 1920, p. 43.
10. ^ Smyth 1920, pp. 42–43.
11. ^ ^{a b c} Allen 1968, p. 116.
12. ^ Smyth 1920, p. 42; Probert 2006, p. 69.
13. ^ Smyth 1920, p. 43; Devine and Stephens 1994, p. 355.
14. ^ Devine and Stephens 1994, p. 355.
15. ^ ^{a b c} Devine and Stephens 1994, p. 475.
16. ^ Devine and Stephens 1994, pp. 475–476.
17. ^ Devine and Stephens 1994, p. 476.
18. ^ Clyde 1857, p. 44.
19. ^ Clyde 1857, p. 52.
20. ^ ^{a b} Devine and Stephens 1994, p. 428.
21. ^ Devine and Stephens 1994, p. 373.
22. ^ Devine and Stephens 1994, pp. 356–357.
23. ^ Devine and Stephens 1994, pp. 359–361.
24. ^ Devine and Stephens 1994, pp. 363–364.
25. ^ ^{a b c} Devine and Stephens 1994, p. 364.
26. ^ Devine and Stephens 1994, pp. 364–365.
27. ^ Clyde 1857, p. 149.
28. ^ Clyde 1857, p. 100.

A noun or adjective keeps its accent on the same syllable through all its forms, as far as the rules allow. This habit is called a persistent accent. To accent any form, you need two things: the accent of the dictionary form, and the length of the last vowel of the form you want.

The rule

In every case, a noun or adjective accents the same syllable as in the nominative singular, if the ultima allows it. If the ultima does not allow it, the accent moves to the next syllable towards the end.^[1] The kind of mark then follows from the rules.

Three things happen again and again when an ending makes the last vowel long:^[2]

1. An acute on the antepenult moves to the penult: θάλαττα, “sea”, genitive θαλάττης; ἄνθρωπος, genitive ἀνθρώπου.
2. A circumflex on the penult becomes an acute: μοῦσα, “muse”, genitive μούσης; δῶρον, “gift”, genitive δώρου.
3. An oxytone of the first or second declension becomes a circumflex in the genitive and dative: θεός, “god”, θεοῦ, θεῷ, θεῶν, θεοῖς.^[3]

Watching the accent move

Here are three nouns of the second declension, one with each possible accent.^[4] In κωμῳδός, “comic singer”, the accent is on the ending and stays there. In παρθένος, “girl”, it is on the penult and stays there. In ἄνθρωπος it is on the antepenult and moves to the penult whenever the ending is long. The red letters show the accented syllable.

The same rules make the circumflex come and go in the first declension. In χώρα, “land”, the α of the penult is long. Before a short final vowel it takes the circumflex, so the plural is χῶραι, because final -αι counts as short. Before a long final vowel it takes the acute, as in χώρας.^[5]

Endings with an accent of their own

Some endings pull the accent onto themselves, whatever the nominative does.

The genitive plural of every first-declension noun ends in -ῶν.^[6] So θάλαττα has θαλαττῶν, and ζώνη, “belt”, has ζωνῶν. Adjectives do not follow this. Their feminine genitive plural has the same accent as the masculine: δίκαιος, “just”, has δικαίων in all genders.^[6]

In the third declension, nouns whose stem has one syllable accent the ending in the genitive and dative of all numbers. The endings -ων and -οιν then take the circumflex.^[7] So θήρ, “wild beast”, has θηρός, θηρί, θηρῶν, θηρσί, and πούς, “foot”, has ποδός, ποδί, ποδῶν, but πόδα in the accusative.^[8] This moving accent is very old. A few words do not accent the genitive dual and plural: παῖς, “child”, has παίδων, but παισί.^[9]

Vocatives

Some vocatives, the forms used to address someone, draw the accent back as far as they can. πατήρ, “father”, has the vocative πάτερ; σωτήρ, “saviour”, has σῶτερ; Ἀπόλλων has Ἄπολλον.^[10] Attic says ἄδελφε, “brother!”, beside the nominative ἀδελφός.^[11] These forms are old survivals and must be learned one by one.^[11] With Ζεύς, “Zeus”, the vocative Ζεῦ moves the high from the second mora of the diphthong to the first.^[12]

Adjectives

Adjectives follow the same persistent rule as nouns. The feminine follows the masculine where it can. In the nominative plural the feminine of an adjective keeps the masculine’s accent, ἄξιαι, “worthy”, like ἄξιοι. A noun of the same shape does not: the noun ἀξία, “value”, has the plural ἀξίαι.^[12] So the accent alone tells these two words apart.^[13]

Notes

1. ^ Smyth 1920, p. 47.
2. ^ Smyth 1920, p. 41.
3. ^ Smyth 1920, pp. 41, 47.
4. ^ Probert 2006, p. 68.
5. ^ Smyth 1920, p. 39.
6. ^ ^{a b} Smyth 1920, p. 47; Probert 2006, p. 66.
7. ^ Smyth 1920, p. 58.
8. ^ Probert 2006, p. 65.
9. ^ Smyth 1920, pp. 58–59.
10. ^ Smyth 1920, p. 63.
11. ^ ^{a b} Probert 2006, p. 67.
12. ^ ^{a b} Probert 2006, p. 66.
13. ^ See Accent and meaning.

Verbs are the easiest words to accent. A finite verb, one with a person and a number such as “I educate” or “they were educating”, has a recessive accent. The accent goes as far back from the end as the rules allow.^[1]

The rule in practice

To accent a finite verb you need only the length of its last vowel. If it is short, the accent goes to the antepenult. If it is long, it goes to the penult. If the verb has only two syllables, the accent stands on the first, and the σωτῆρα rule chooses between acute and circumflex.

So παιδεύω, “I educate”, has its accent on the penult, because the final ω is long. παιδεύομεν, “we educate”, has it on the antepenult, because the final ε is short. The accent moves from form to form only because the endings differ in length.^[2]

The augment, the ἐ- that marks a past tense, gives the accent one more syllable to go back to. So the imperfect is ἐπαίδευον, “I was educating”, with the accent on the antepenult. In ἐπαιδεύομεν, “we were educating”, it is again on the antepenult, which is now a different syllable.

Final -αι counts as short here too, except in the optative.^[3] So παιδεύομαι, “I am educated”, is accented on the antepenult. The aorist of παιδεύω shows all three possibilities in one spelling.^[4] παιδεύσαι, “may he educate”, is an optative, so its -αι is long. παιδεῦσαι, “to educate”, is an infinitive, and infinitives are not recessive. παίδευσαι, “educate yourself!”, is an imperative, whose -αι is short, so it is recessive.

Why verbs are recessive

The Greek verb was recessive in every dialect we know, so the habit is very old.^[5] A widely accepted explanation goes back to the Vedic Sanskrit of ancient India. There, a finite verb was enclitic in certain positions in the sentence, with no accent of its own. When Greek limited the accent to the end of the word, a long verb could no longer be fully enclitic. Instead it took an accent as far back as it could, and this became the rule for all finite verbs.^[5]

What is not recessive

Infinitives and participles are nouns and adjectives made from verbs. They do not follow the verb’s rule:^[2]

• Infinitives in -ναι accent the penult: λελυκέναι, “to have loosed”, λυθῆναι, “to be loosed”. So does the first aorist infinitive active, παιδεῦσαι.
• Some participles are oxytone: λιπών, “having left”, λυθείς, “having been loosed”. Once a participle has its accent, it keeps it like an adjective: παιδεύων, παιδεύουσα, παιδεῦον.^[3]

There are a few other exceptions:^[2]

• The present indicative of φημί and εἰμί is enclitic.^[6]
• Five short imperatives are oxytone: εἰπέ, “say!”, ἐλθέ, “come!”, εὑρέ, “find!”, ἰδέ, “look!”, and λαβέ, “take!”.
• In a compound verb the accent cannot go back past the augment. So the past tense of ἄπειμι, “I am away”, is ἀπῆν, “was away”, not *ἄπην.^[3]

Contracted verbs

Many verbs have stems ending in α, ε or ο. In Attic this vowel merges with the vowel of the ending, as φιλέω, “I love”, becomes φιλῶ. These verbs are recessive too, but the accent is placed before the vowels merge.^[2] Then the merged vowel takes the accent by a simple rule.^[7]

If one of the two vowels was accented, the merged vowel is accented. It takes a circumflex if the accent was on the first of the two vowels, because the high then falls on the second mora. It takes an acute if the accent was on the second.^[7] On the penult the σωτῆρα rule applies as usual. If neither vowel was accented, the merged vowel has no accent: φίλεε becomes φίλει, “love!”.^[7]

Notes

1. ^ Smyth 1920, p. 143; Probert 2006, p. 86.
2. ^ ^{a b c d} Smyth 1920, p. 143.
3. ^ ^{a b c} Smyth 1920, p. 144.
4. ^ Probert 2006, p. 61.
5. ^ ^{a b} Probert 2006, p. 87.
6. ^ See Enclitics and proclitics.
7. ^ ^{a b c} Smyth 1920, p. 40.

The accent is part of a word, like its vowels and consonants. Change it, and you may say a different word. Speakers of Greek did not need accent marks to tell such words apart, because they heard the difference, and the sentence usually made the meaning plain.^[1] In writing, though, the words looked the same.

The ancients noticed

Long before the marks existed, Greek writers noticed what the accent did. In Plato’s Cratylus Socrates observes that the phrase Διὶ φίλος, “dear to Zeus”, becomes the name Δίφιλος by losing a letter and changing its accent.^[2]

Aristotle collected puzzles that only work on paper. One turns on οὐ, “not”, and οὗ, “where”: “Is where you lodge a house? And is you don’t lodge the negation of you lodge? Then a house is a negation.” The answer, he says, is that the word does not mean the same when it is said on a higher pitch as on a lower.^[3] Others used the same pair in Homer. At Iliad 23.328 a wooden post either does not rot in the rain, οὐ καταπύθεται, or rots in part, τὸ μὲν οὗ καταπύθεται.^[2]

The most famous slip was made on the stage. The actor Hegelochus, in Euripides’ Orestes, meant to say γαλήν᾽ ὁρῶ, “I see a calm”, but was heard to say γαλῆν ὁρῶ, “I see a weasel”. The comic poets made fun of him for it.^[4]

Later writers show that the accent still mattered when a text was read aloud. The rhetorician Hermogenes imagines a law about a courtesan who steals gold pieces. The law says that they shall be ΔΗΜΟΣΙΑ, “public property”. Read as δημόσια, the word means the gold pieces. Read as δημοσίᾱ, it means the woman herself.^[5] The doctor Galen could not decide whether a rare word in Hippocrates was γαῦσος or γαυσός, “crooked”, and told his readers to say it as their listener did.^[6]

Pairs of words

These pairs differ only in their accent. Press each button and listen for where the voice is high.

Forms of one word

Inside one word’s paradigm, the accent often tells forms apart.

• οἶκοι, “houses”, is a plural, so its final -οι counts as short and the long penult takes a circumflex. οἴκοι, “at home”, is an old locative, whose -οι counts as long, so the penult takes an acute.^[7]
• παιδεύσαι is an optative, παιδεῦσαι an infinitive and παίδευσαι an imperative.^[8] In the same way λῆξαι, “to cease”, differs from λήξαι, “may he cease”.^[9]
• Ζεύς, “Zeus”, has its high on the second mora of ευ; the vocative Ζεῦ has it on the first.^[10]
• ἄξιαι, “worthy” (feminine plural), keeps the masculine’s accent; ἀξίαι, “values”, is a noun.^[10]
• βασίλεια, “queen”, has a short final α; βασιλεία, “kingdom”, has a long one, so its accent cannot go back to the antepenult.^[11]
• τίς, “who?”, is always accented; τις, “someone”, is enclitic. The same holds for πότε, “when?”, and ποτε, “ever”, and for ποῦ, “where?”, and που, “somewhere”.^[12]

The Practice tab has a listening test with these pairs.

Notes

1. ^ Allen 1968, p. 114.
2. ^ ^{a b} Probert 2006, p. 16.
3. ^ Probert 2006, p. 17; Allen 1973, p. 253.
4. ^ Probert 2006, p. 18; see Pitch accent.
5. ^ Probert 2006, p. 20.
6. ^ Probert 2006, pp. 19–20.
7. ^ Smyth 1920, p. 39; Probert 2006, p. 61; Allen 1973, p. 91.
8. ^ Probert 2006, p. 61.
9. ^ Allen 1968, p. 115.
10. ^ ^{a b} Probert 2006, p. 66.
11. ^ Smyth 1920, p. 39.
12. ^ Smyth 1920, p. 42.

Each exercise gives an explanation as soon as you choose, whether you are right or not. Your score counts only the questions you got right at the first try. Nothing is saved: reload the page to start afresh.

A song, word by word

The epitaph of Seikilos is a short song inscribed on a tombstone found at Aidin, near ancient Tralles in Asia Minor. It is not earlier than the second century BC, and probably of the first century AD.^[1] Its melody follows the word accents closely, which makes it a good text to hear them in.^[2] Press a word to see its morae and hear it said on its own. In the line itself, a grave stands for the acute of a word that another word follows.

As long as you live, shine; let nothing grieve you at all. Life lasts only a little while, and time asks for its end.

Which did you hear?

Each clip says one word of a pair that differs only in its accent. Listen, then choose.

Acute or circumflex?

The accented syllable is underlined, and long α, ι and υ carry a macron. Choose the mark.

Accent the verb

Finite verbs are recessive. Choose the correctly accented form.

Move the accent through a noun

The nominative is given. Choose the correct form of the case asked for.

Words that lean

Choose the correct accents for the host and its enclitic.

Notes

1. ^ Allen 1968, p. 109.
2. ^ Allen 1968, pp. 109, 112.