Ancient Greek Accents in Ten Rules

David Butterfield

Pity poor Hegelochus. This Ancient Greek actor once provoked unexpected laughter during the premier of the Euripidean tragedy Orestes in 408 BC. What was so funny? In a lapse of concentration, he unwittingly delivered a line where the expected γαλήν’ (elided γαληνά) came out as γαλῆν (accusative of γαλῆ). So the crowd heard the distressed Orestes announce to his sister and the world, not “after the storm I see again the calm sky”, but “after the storm I see again the… weasel”.[1] Madness, they say, can take many forms.

The Aegean at rest, 408 BC (colourised).

Just like the Furies, Ancient Greek accents have been known to drive students mad because of their apparent complexity. But this need not be so: while these strange squiggles seem baffling at first view, they are in reality quite well behaved. So, if you are a reader of Greek texts and are still to get full control of these diacritical marks, Antigone offers you this (relatively) short list of ten rules.[2]

Before we come to them, let us set out some simple truths. Unlike languages such as English, Hungarian or Latin, which have a stress accent that raises the volume and emphasis on particular syllables,[3] Ancient Greek modulated not its stress but its tonal pitch: as in Punjabi or Mandarin Chinese,[4] the voice was raised and lowered on particular syllables.[5] Changes in pitch could change meaning: the word for “houses” was οἶκοι, with a rise and fall in pitch on the first syllable; but if you only raised the pitch on that first syllable – οἴκοι – you instead expressed the sense “at home”. And if you saw the letters παιδευσαι, you would naturally want to know whether the intended intonation was παίδευσαι (“get (someone) educated”), παιδεύσαι (“he/she should educate”) or παιδεῦσαι (“to educate”).

Thomas Warton’s ill-fated Oxford Theocritus (1770), which tried printing Greek without smooth breathings or accents, alongside Heinrich Ahrens’ Teubner edition (Leipzig, 1855).

It is helpful, then, that Greek accents are written out before our very eyes: they stand above the vowels that form the heart of every syllable.[6] First introduced in the second century BC by the grammarian Aristophanes of Byzantium (and soon refined by the celebrated scholar Aristarchus of Samothrace), written accents help readers of Classical Greek texts pronounce words and sentences as they should be.

Aristarchus wins an invitation to Jean-Auguste-Dominique Ingres’ Apotheosis of Homer, 1827 (Louvre Museum, Paris, France).

So let’s embrace the considerable help that these subtle scratches of the pen can offer and – with Attic Greek as our primary guide – let’s take a closer look at the ten rules that matter.

1) There are three accents, all of which involve the raising of vocal pitch of the vowels they stand over, here exemplified by α: an acute (/, written as ά), which means the pitch rises; a circumflex (~, originally /\, written as ᾶ), which means the pitch rises and then falls; and a grave (\, written as ὰ), which means the syllable’s pitch rises, but to a lesser degree than in the other two cases.[7] On all other (unmarked) syllables, the pitch is at its lower, neutral tone.

This is enough to know to appreciate the rise and fall of any piece of written or spoken Ancient Greek. The rest of the rules explain how it is that certain accents appear in certain positions and never others, and how some rather strange things can happen when particular words are combined.

2) Each word should carry one accent. If you find a word that has no accent marked, or a word that has two accents at different places, what you are really seeing is a word that has been closely paired with another word. There are, you see, eleven words that are pronounced so closely with the word that follows that they don’t have sufficient autonomy to possess their own accent. These are: four forms of the definite article: the masculine and feminine singular and plural nominatives ὁ, ἡ, οἱ, αἱ;[8] four prepositions: ἐκ / ἐξ (“from”), εἰς / ἐς (‘to”), ἐν (“in”), ὡς (“to”); two conjunctions: εἰ (“if”), ὡς (“as”, “that”, etc.); and one adverb: οὐ / οὐκ / οὐχ (“not”).

3) Further, each of these accents is differently restricted in its placement. An acute may stand on any of the last three syllables of a word. In fact, it is much more commonly found on the antepenultimate (e.g. κράτιστος) or penultimate (κράτος) syllables than on the last (κακός):[9] unless a word precedes punctuation, and thus has a natural pause following it, an acute on the final syllable is replaced by a grave (e.g. κακὸς βοῦς).[10]. Since this grave accent denotes the lower rise of pitch found exclusively on a word’s last syllable, it can only ever be found there. Whereas acutes and graves can stand on short or long vowels, a circumflex, denoting a rise and then a fall, can only stand above a long vowel (or diphthong), which gives it sufficient time to change pitch twice. It can stand on the penultimate or last syllable of words.[11] A reminder, then: acute accents can stand on any of the last three syllables, circumflex accents on either of the last two, grave only on the last.

4) A further restriction exists for the placement of the acute and the circumflex: if the vowel or dipthong of the last syllable of a word is long,[12] an acute accent cannot stand any earlier than the penultimate syllable, and a circumflex can only be found on the last (in this case long) syllable of such a word.[13] In practice, this means that an acute accent can be moved from the antepenultimate to the penultimate syllable, when the last syllable changes its quantity from short to long: ἄνθρωπος in the nominative becomes ἀνθρώπου in the genitive. For a circumflex that stood on the penultimate syllable, it is converted to an acute, when the last syllable changes its quantity from short to long: τοῦτο in the nominative becomes τούτου in the genitive.

But just before you roll on to the next rule, here’s a curious point to note: for the purposes of accentuation, the nominative (and vocative) plural endings –οι and –αι in nouns and adjectives, along with the ending –αι in finite verbs and infinitives, are treated as short, although they are metrically long in poetry: ἄνθρωποι, μοῦσαι, λέγονται. These endings, which are as common as could be, were perhaps said with less emphasis – and therefore spoken length – in standard Greek.

5) At this point a recurrent phenomenon should be noted: if a word’s accent is due to stand on its penultimate syllable, and if that syllable has a long vowel or diphthong, its accent must be a circumflex, provided that the word ends with a short syllable. In other words, if a word ends with a long and then a short syllable (– ᴗ, the “trochee” rhythm), and if the accent should stand on the penultimate, it must be a circumflex: παύει but παῦε, ταύτη but οὗτος.[14]

6) In nouns, the accent stays in the same position as its place in the nominative singular, despite changes to its number and case;[15] in adjectives, it stays in the same position as in the neuter nominative singular, despite changes to its number, case and gender. The effect of final syllables becoming long (as per Rule 4) still applies.[16] There are three exceptions to this rule, one for each declension.

    • The genitive plural of first-declension nouns (and many adjectives) always ends with a circumflex on the final –ῶν, because it represents the contraction of original –άων.
    • Nouns and adjectives of the second declension, if accented on their last syllable, change their acute accents to circumflexes in the genitive and dative cases, singular and plural: nominative ὁδός and accusative ὁδόν but genitive ὁδοῦ and dative ὁδῷ (so in the plural: ὁδοί, ὁδούς but ὁδῶν, ὁδοῖς).
    • Monosyllabic nouns of the third declension place the accent on their latter (final) syllable in the genitive and dative cases: nominative χείρ and thus accusative χεῖρα (as per Rule 5), but genitive χειρός and dative χειρί; likewise in the plural we have nominative χεῖρες and thus accusative χεῖρας, but genitive χειρῶν and dative χερσί.

7) Verbs like to place their accent as far back as the rules above allow.[17] So we find, for instance, ἔτυπτον, φυλάττειν, λέγονται (as per Rule 4) and δρᾶσον (as per Rule 5). But there are certain parts of the verbal system where this rule is overturned, and the accent occurs in a later position:

    • Strong aorist active infinitives have final circumflexes: λαβεῖν, δραμεῖν.
    • Weak aorist active, perfect active, strong aorist middle, perfect middle and perfect passive infinitives have their accent on the penultimate syllable, and follow the ‘trochee rule’ (Rule 5) where necessary: καλέσαι, τιμῆσαι, λελυκέναι, λαβέσθαι, λελύσθαι, πεποιῆσθαι. This exception also applies to all infinitives ending in –ναι: διδόναι, ἰέναι, στῆναι, etc.
    • Strong aorist active participles, all participles in –εις and –ως, and all active participles from –μι verbs have a final acute in the masculine nominative singular: λαβών, λυθείς, λελυκώς, ἱστάς, and the accent remains on this syllable in other forms (e.g. λαβόντος, λαβοῦσα etc.).
    • Perfect middle / passive participles always have an acute on the penultimate: λελυμένος as well as λελυμένου, πεποιημένος as well as πεποιημένοις.
    • All strong aorist middle 2nd singular imperatives have final circumflexes: λιποῦ, λαβοῦ etc.[18]
    • Aorist passive subjunctives have circumflexes on the crucial –(θ)ῶ- syllable: λυθῶ, λυθῶμεν, λυθῶσι.

8) If two adjacent vowels are contracted into one syllable, the accent will become a circumflex, if the accent previously stood on the earlier syllable (νόος contracts to νοῦς,[19] but remains as an acute if it stood on the latter syllable of the contracted pair (χροός contracts to χρώς). The accentuation of contract verbs (in –άω –έω –όω, usually appearing as –ῶ) also follows this rule, contracting the normal accentuation of the pre-contracted verb: ποιοῦμεν represents the contraction of ποιέομεν, and τιμῷεν the contraction of τιμάοιεν.[20] Similarly, we can distinguish between ποίει from the imperative ποίεε (where the contracted syllable never had any accent) and ποιεῖ from indicative ποιέει.

9) What about when the syllable that should have an accent is lost in elision? In Greek, it is only short vowels that allow elision, so we’re not having to deal with circumflexes going AWOL. In practice, we are talking about word-final grave accents being lost in the fray. The answer is simple enough: if the word is declinable, the accent moves back to stand as an acute on the previous (unelided) syllable:[21] elide πολλὰ before ἐθέλω and the outcome will be πόλλ’ ἐθέλω. If the word is indeclinable and/or monosyllabic, the accent disappears without a trace: ἀλλὰ οὖν and ἀπὸ ἐμοῦ become merely ἀλλ’ οὖν and ἀπ’ ἐμοῦ.

Well, we are nearly there now, but the tenth and final rule will keep you on your toes:

10) Some words in Greek cohere so closely with the word that precedes that they are pronounced as part of the same unit.[22] We call these words enclitics.[23] In practice, this means that they affect the accent of the preceding word, and often share only one accent between them. If that word should have a grave, since another word follows it, a subsequent enclitic instead allows its pitch to rise properly and be an acute: σοφὸς ὗς but σοφός τις. If instead, the word had an acute on the antepenultimate, or a circumflex on the penultimate, because the pitch has fallen again by the last syllable, it is able to rise again under the influence of the enclitic: in both of these circumstances – and only these – words can bear two accents: ἄνθρωπός τις and τοῦτό γε. Yet if an enclitic follows a word which has a circumflex on its final syllable, that preceding word is unchanged and the enclitic carries no accent: τιμῶ σε and καλῶς φημι. Finally, if the preceding word has an acute on its penultimate syllable, that word is unchanged, and any monosyllabic enclitic is unaccented; but disyllabic enclitics will bear an acute accent on their last syllable [24], which will become the expected grave under the influence of a following word: λόγος τις and τύχῃ τινὶ ἀπέθανε. No monosyllabic enclitic word bears its own accent.[25]

Demosthenes practising oratory, Jean-Jules-Antoine Lecomte du Nouÿ, 1870 (private collection).

These rules should give you sufficient working knowledge of what will happen to the accents of words within any given sentence. In order to know where the accent should be placed to begin with in nouns, adjectives, adverbs and the like, that is something learned on the job. The dictionary will give you the right answer when you are not sure, and the more you learn, the more you will see patterns about where the default accent tended to be placed. You will come to know why we find καλός but κάλλιστος; odd words that seem μυρίοι in number will emerge to be far fewer than μύριοι; but ὀλίγα curiosities are πολλάκις καταληπτέα.[26]

So go forth with good luck. For much more information, along with some exercises and bibliographical leads, the best starting place is Philomen Probert’s New Short Guide to the Accentuation of Greek (Bloomsbury, London, 2003). Oh, and if you think there really should be another rule finding its place among these ten, please do drop us a line.


1 Line 279, and the passage in which it occurs, can be read in Greek and English here; the source for this famous story is a scholion – or ancient scholar’s commentary – on the garbled line. The comic playwright Aristophanes, never one to miss a jibe at Euripides, mocked the event three years later in his Frogs, line 303, readable here.
2 OK, this is one of 26 footnotes…
3 The rules for English stress are very complex (doable, hairdo, ado?!), in Latin they are pretty simple (atus and cona, but gĭus and fundĕre, more on this at p.2 here), and in Hungarian they are even simpler (the first syllable: egészségedre! “cheers!”, viszontlátásra “goodbye”).
4 In Punjabi, ਕਰ kar is “do”, but the same sound with a falling pitch is ਘਰ kàr (“house”), and with a rising pitch ਕਰ੍ਹ kár (“dandruff”). In Mandarin, 麻  with a rising tone means “cannabis”, but 马 , with a fall and then rise in pitch means “horse”.
5 Over time, Ancient Greek evolved so as to have a stress accent, although signs of this do not start appearing until the 2nd century AD.
6 When more than one vowel occurs in the syllable, the following rule applies: place the accent above the second vowel of the diphthong, unless it is the iota adscript, in which case the preceding vowel bears the accent. If an accent stands on the initial vowel of a word, acutes are written after the breathing, circumflexes spread over and above the breathing like Athena’s aegis. If the initial vowel is capitalised, by convention the accent is written with the breathing to the left of the capital letter. So we find ἔθος and ἦθος and ᾍδης
7 You won’t be surprised that there are more technical terms for these accents: the acute is the oxytone, the circumflex the perispomenon, and the grave the barytone.
8 The remaining forms of the article have acute accents (nearly always in practice made grave, according to Rule 3) in the nominative and accusative, but circumflexes in the genitive and dative: neuter nominative singular τό, accusative singular τόν τήν τό, and neuter nominative plural τά, accusative plural τούς τάς τά, but genitive singular τοῦ τῆς τοῦ, dative singular τῷ τῇ τῷ, genitive plural τῶν τῶν τῶν, dative plural τοῖς ταῖς τοῖς.
9 The accent has some further technical names when not found on the last syllable: proparoxytone for an acute on the antepenultimate, and paroxytone on the penultimate, leaving plain old oxytone for its placement on the last.
10 Rules wouldn’t be rules without exceptions, so here’s one to remember: the interrogative pronoun and pronominal adjective τίς is the only word that never changes its acute accent to a grave under the influence of a following word: e.g. τί λέγεις;.
11 Again, some technical terms for you: properispomenon for a circumflex on the penultimate syllable, alongside mere perispomenon for a circumflex on the last.
12 When talking about ‘short’ and ‘long’ vowels in the context of accentuation, we are speaking only of the vowels, not the metrical quantity of the syllable they occupy (on which see pp.3-5 here). So a word such as πολλά has two ‘short’ syllables for accentual purposes, even though πολλ- would always ‘scan’ as long in verse.
13 Another exception: there are a few forms, such as the genitive singular of πόλις, the word for “city”, i.e. πόλεωϛ, that have an acute on the antepenultimate despite having a long vowel in their last syllable. The reason for this has to do with a process known as quantitative metathesis (μετάθεσιϛ “transposition”): the sequence -εω-, with a short vowel followed by a long, used to be -ηο-, with a long vowel followed by a short (so πόληοϛ was the legitimate accent position, which remained where it was after metathesis).
14 Apparent exceptions to this rule, such as ἥδε, οἵπερ, οὔτε, ὥστε, καίτοι, were originally two words, a history that their accents (i.e. their pronunciation) still reflect. This anomaly can be found more generally in forms of ὅστις and ὅδε, which are accentuated as if the τις and δε parts were absent: so we find ἥτις and ἥδε (not the unattested ἧτις / ἧδε), and apparently illegal circumflexes (according to Rule 4), such as οὗτινος and ὧντινων.
15 Striking exceptions are rare, but an infamous one is given by the word for “brother”, ἀδελφός, which throws its accent to the start of the word in the vocative ἄδελφε. Strange but true.
16 At this point you may ask, “Well where does it appear in the nominative singular?” That is a perfectly fair question, and one that the dictionary will often have to tell you the answer.
17 Such a placement is termed recessive.
18 NB: Five strong aorist active 2nd singular imperatives from particularly common verbs have an anomalous final acute: εἰπέ, ἐλθέ, εὑρέ, ἰδέ, λαβέ, although compounds of these verbs become regular, e.g. ἄπελθε.
19 Cf. genitive plural –άων becoming –ῶν in Rule 6.
20 This rule does not apply for those tenses not based upon the present-tense stem: forms with the lengthened stem (such as τιμη- / ποιη- / δηλω-) have the expected recessive accent.
21 If you are in search of an exception, enclitic τινά is the word for you.
22 The most common of these are the indefinite pronoun τις (all forms), the indefinite adverbs ποτε πως ποθεν πω πῃ ποι που ποθι, the particles τε γε περ τοι νυν (and Homeric κε νυ ῥα), the (unemphatic) personal pronouns με μου μοι, σε σου σοι, ἑ οὑ οἱ, σφε σφεων σφι(σι), along with poetic μιν and νιν, and the present indicative of (non-existential) εἰμι and φημι, excepting the second singulars εἶ and φῄς.
23 It should be noted that Greek also has a class of words which must follow other words but which are not strictly enclitic: these so-called postpositives, including very common words such as δέ, γάρ and μέν, conventionally possess their own accent.
24 NB however τινοῖν and τινῶν.
25 If you see an instance where it seems to, it will actually be bearing an accent imposed by another following enclitic: a string of enclitic words can all end up ‘passing’ their accent backwards to the preceding word, e.g. εἴ τίς ποτέ φησί σοί γε τοῦτό πω. If a proclitic precedes an enclitic – what a meeting! – the proclitic is able to inherit an accent from the enclitic, e.g. ἥ τε πόλις, “and the city”.
26 Readers may be interested that Chandler’s Conversion – an idea for a musical that charts a moral philosophy professor’s internal struggle over the ethics of Greek accentuation – is currently seeking crowdfunding. To whet the appetite, compare the prefaces to his first edition of A Practical Introduction to Greek Accentuation (Oxford, 1862) with the second (Oxford, 1881): “Whether a skilful advocate could convince, I do not say a mere verbal scholar, for that would be easy, but a man of sense, that a knowledge of the subject is worth the time and trouble which must be expended to acquire it, may or may not be doubtful, but it is certain that for the present all who pretend to a critical knowledge of the Greek language must yield perforce to a tyrannous custom, or refusing to do so, must expect to be rebuked for their ignorance by those who are unable to see the absurdity of perpetuating in writing a something to which they never attend in reading, and who persist in ornamenting their Greek with three small scratches, the very meaning of which is doubtful and perhaps unknown.”  Twenty years on, he wrote:  “In bidding a last farewell to a subject in which I never took more than a languid interest, I may be permitted to say that in England, at all events, every man will accent his Greek properly who wishes to stand well with the world. He whose accents are irreproachable may indeed be no better than a heathen, but concerning that man who misplaces them, or, worse still, altogether omits them, damaging inferences will certainly be drawn, and in most instances with justice.”