Mova0.6 pre-release⌕
☰
home / docs / languages / grc / Accent

Accent

Greek accent in the fifth century BC was a pitch accent. The classical texts were written without accent marks; the marks were invented by the Alexandrian scholars (Aristophanes of Byzantium, about 200 BC, is credited with the system) and fixed by the grammarians of the imperial age (Herodian, second century AD). Everything an edition prints today is therefore a scholarly reconstruction of an oral accent, but it is a very stable one, and it is a part of the identity of a Greek word: it tells the quantity of a vowel, separates words that are otherwise spelt alike, and is part of the data a morphological analyser must carry.

This chapter gives the three marks, the limits on their position, the accent of nouns and verbs, the clitics, and the changes that contraction, crasis, elision and composition cause. Section references are to Smyth.

Contents

  1. The marks and the terms
  2. The limits: where an accent may stand
  3. Accent on nouns and adjectives
  4. Accent on verbs
  5. Accent after contraction, crasis and elision
  6. Proclitics
  7. Enclitics
  8. Anastrophe and the grave
  9. Accent that distinguishes words
  10. What accent was, and what became of it
  11. Worked examples
  12. How Mova handles it
  13. Sources

1. The marks and the terms

marknamewhereexample
´acute (ὀξεῖα)on a short or long vowel, last three syllablesλόγος, ἄνθρωπος, ποιητής
`grave (βαρεῖα)only on the last syllable, in place of an acute, before another wordὁ ἀγαθὸς ἀνήρ
῀circumflex (περισπωμένη)only on a long vowel or diphthong, last two syllablesσῶμα, τιμῶ

Words are named by the place and mark of the accent (smyth-1920 §153):

The grave is a convention of the writing: the last syllable of a word that would be oxytone in isolation is written with a grave when another word follows without a pause (καὶ ἀγαθὸς ἄνθρωπος, but ἄνθρωπος ἀγαθός at the end of a clause). Interrogatives (τίς, τί) keep their acute.

2. The limits: where an accent may stand

The accent may stand only on one of the last three syllables of a word. Beyond that, two facts about the final syllable decide (smyth-1920 §§159–170):

The quantity is that of the vowel as in chapter 1, with one special rule: final -αι and -οι count as short (μοῦσαι, ἄνθρωποι, βούλομαι), except in the optative (λύσαι, βουλεύοι) and in the locative οἴκοι "at home" (against οἶκοι "houses") (smyth-1920 §169). Two further consequences:

3. Accent on nouns and adjectives

The persistent accent is the rule: the accent of a noun stays on the syllable it occupied in the nominative singular as long as the limits permit, and the mark changes only as the quantity of the ending requires. So, from an accent of the nominative: ἄνθρωπος, ἀνθρώπου, ἀνθρώπῳ, ἄνθρωπον, ἄνθρωποι, ἀνθρώπων; θάλαττα, θαλάττης. In the second declension the oxytone becomes perispomenon in the genitive and dative: θεός, θεοῦ, θεῷ, θεῶν, θεοῖς (§176c). The third declension has a number of fixed groups:

The accent of the adjective in its agreement follows the same persistence: ἀγαθός, ἀγαθή, ἀγαθόν; the participles and the third declension adjectives show their own patterns (chapters 4 and 10).

4. Accent on verbs

Verbs have recessive accent: the accent goes as far from the end of the word as the limits allow. The long or short quantity of the ultima fixes the place: λύω (long ultima → penult), λύομεν (short -εν → antepenult), λύει, ἔλυσα, ἐλύσαμεν, ἔλυσαν. The augment counts as a syllable: a form with an augment can be accented on it (ἔλυον, ἔπεμπον, ἔλειπον) and the accent cannot recede beyond the augment.

Exceptions to the recessive rule (smyth-1920 §§423–427):

Contract verbs and those of the -μι class have their own accent and are described where the forms are (chapters 8 and 9).

5. Accent after contraction, crasis and elision

Contraction (smyth-1920 §§171–172). If either of two syllables that contract had an accent, the contracted syllable takes one. A contracted antepenult has the acute (φιλεόμενος → φιλούμενος); a contracted penult has the circumflex when the ultima is short, the acute when the ultima is long (φιλέουσι → φιλοῦσι, φιλεόντων → φιλούντων). A contracted ultima has the acute when the uncontracted form was oxytone (ἑσταώς → ἑστώς), and the circumflex otherwise (φιλέω → φιλῶ). If neither syllable had the accent, the contracted syllable has none (φίλεε → φίλει). The ancient grammarians' explanation is that an acute followed by the underlying fall (grave) on the next mora gives the circumflex: the circumflex is the result of a rise-fall on one syllable.

Crasis (§173). The first word loses its accent: τἀγαθά (τὰ ἀγαθά), κἀγώ (καὶ ἐγώ).

Elision (§174). Oxytone prepositions and conjunctions lose their accent: παρ᾽ ἐμοῦ, ἀλλ᾽ ἐγώ. In other oxytones the accent is thrown back as an acute on the previous syllable (πόλλ᾽ ἔπαθον for πολλὰ ἔπαθον, φήμ᾽ ἐγώ for φημὶ ἐγώ). A circumflex does not result from this recession.

6. Proclitics

Ten monosyllabic words have no accent of their own and lean on the following word (smyth-1920 §179): the forms of the article that begin with a vowel (ὁ, ἡ, οἱ, αἱ); the prepositions ἐν, εἰς (ἐς), ἐξ (ἐκ); the conjunction εἰ "if"; ὡς "as, that, to"; and the negative οὐ (οὐκ, οὐχ). A proclitic takes an accent in a few cases (§180): οὐ at the end of a sentence (ἢ οὔ; "or not?"), ὡς in the sense of οὕτως (ὥς), a proclitic before an enclitic (ἔν τινι), ὥς in poetry after its noun.

7. Enclitics

Enclitics lean on the preceding word and usually lose their own accent (smyth-1920 §§181–187). They are:

The accent of the word before the enclitic (§183):

the preceding wordthe ruleexample
oxytone or perispomenonunchanged, the enclitic loses its accentδός μοι, φιλῶ σε
proparoxytone or properispomenontakes an additional acute on its last syllableἄνθρωπός τις, ἄνθρωποί τινες, παῖδές τινες
paroxytoneno extra accent; a monosyllabic enclitic loses its accent, a dissyllabic one keeps itφίλος μου, χώρας τινός, φίλοι τινές
a proclitictakes an acuteἔν τινι, εἴ τινες

When several enclitics follow each other, each takes an accent from the next and only the last has none: εἴ πού τίς τινα ἴδοι ἐχθρόν "if anyone anywhere saw an enemy" (§185, from Thucydides).

The enclitic is orthotone (keeps its accent) in these conditions (§187): when it is emphatic, as in a contrast (ἢ σοὶ ἢ τῷ πατρί σου), or at the beginning of a clause (φημὶ γάρ); after a word that has suffered elision (πολλοὶ δ᾽ εἰσίν); when a dissyllabic enclitic follows a paroxytone (φίλοι τινές); and, among the oxytone prepositions, after them (ἐπὶ σοί, not ἐπί σοι). The form ἐστί is written ἔστι at the beginning of a sentence, when it expresses existence or possibility, and after οὐκ, μή, εἰ, ὡς, καί, ἀλλά and τοῦτο (εἰ ἔστιν οὕτως "if it is so"). The difference between existential ἔστι and copulative ἐστί is thus visible in the accent only. Sentence 1.2.8 of the Anabasis uses both:

An. 1.2.8 · tlg0032.tlg006 ἔστι δὲ καὶ μεγάλου βασιλέως βασίλεια ἐν Κελαιναῖς … τοῦ δὲ Μαρσύου τὸ εὖρός ἐστιν εἴκοσι καὶ πέντε ποδῶν "there is also a palace of the Great King at Celaenae … and the width of the Marsyas is twenty-five feet"

ἔστι begins the clause and is existential: it keeps its accent. ἐστιν follows εὖρός: the properispomenon εὖρος takes an additional acute on its last syllable, and the enclitic loses its own accent (§183c). The final ν is the movable ν of chapter 1.

8. Anastrophe and the grave

Anastrophe (ἀναστροφή) throws the accent of a disyllabic preposition back to the first syllable when it follows its noun: τούτων πέρι "about these things" for περὶ τούτων. In Attic prose only περί follows its case; in poetry the other disyllabic prepositions do too, except ἀντί, ἀμφί, διά (smyth-1920 §175). In Homer the same occurs with a preposition after its verb in tmesis (λούσῃ ἄπο for ἀπολούσῃ). The compound πάρα (πάρεστι "it is possible") and ἔνι (ἔνεστι "there is in it") are anastrophe of a preposition for a compound with εἰμί.

The grave (§154): the acute on the ultima becomes a grave when the word is followed immediately, without punctuation, by another word that is not an enclitic. In editions this appears everywhere (καὶ, πρὸς, ἀγαθὸς), and the oxytone character of a word is then visible only from its isolated form.

9. Accent that distinguishes words

Accent separates words that spelling alone would confuse. This is the practical reason why accents cannot be dropped in a Greek tokeniser:

pairmeaning
θεά / θέαgoddess / sight
βίος / βιόςlife / bow
ἡ / ἥ / ἤthe (fem.) / who (fem.) / or, than
ἦ / ἢtruly (also a verb form) / or
εἰ / εἶif / you are
οὐ / οὗnot / of whom, where
τίς / τιςwho? / someone
ποῦ / πουwhere? / somewhere
πῶς / πωςhow? / somehow
πότε / ποτέwhen? / once, ever
ὅς / ὁ / ὅ / οἷοςwho / the / which / such as
ἔστι / ἐστίthere is / is
εἶμι / εἰμίI shall go / I am
ἔσται / ἐστέwill be / you are

10. What accent was, and what became of it

The ancient grammarians describe the acute as a rise of about a musical fifth on one syllable, the circumflex as a rise and fall on one long syllable, and the grave as the absence of the rise. The accent on the last syllable fell to a lower pitch when other words followed (the grave). By the second to third centuries AD, in the Koine, the pitch accent became a stress accent and the vowel quantity disappeared as a distinction; this is why the accent marks of Byzantine copyists have no effect on pronunciation in Modern Greek except to place the stress, and why metrical poetry of the late empire is written by counting accents (manual grc-x-koine, chapter 2).

11. Worked examples

Il. 1.1 · tlg0012.tlg001 μῆνιν ἄειδε θεὰ Πηληϊάδεω Ἀχιλῆος "sing, goddess, the wrath of Achilles son of Peleus"

μῆνιν: properispomenon (long penult, short ultima: circumflex). ἄειδε: imperative, recessive, on the first syllable. θεὰ: oxytone θεά, grave before the next word. Πηληϊάδεω: paroxytone, long ultima -ω (Ionic genitive) forces the acute on the penult. Ἀχιλῆος: circumflex on the long penult, since the ultima -ος is short.

Pl. Ap. 17a · tlg0059.tlg002 ὅτι μὲν ὑμεῖς, ὦ ἄνδρες Ἀθηναῖοι, πεπόνθατε ὑπὸ τῶν ἐμῶν κατηγόρων, οὐκ οἶδα "what effect my accusers have had on you, men of Athens, I do not know"

ὅτι is paroxytone (acute on the first syllable). μὲν: oxytone-grave before another word. πεπόνθατε: proparoxytone (perfect indicative; the ultima -τε is short). ὑπὸ: grave of ὑπό. κατηγόρων: paroxytone with long ultima. οὐκ: proclitic, no accent. οἶδα: properispomenon (long penult, short ultima).

12. How Mova handles it

treebankτίς "who?" and τις "someone"
PTNKone lemma τίς, PRON, PronType=Int, even for the unaccented τις (all 7 occurrences of unaccented τις are Int)
Perseuslemmas τις and τίς both used for both words (τί is lemma τις 88 times, τίς 130 times); no PronType
PROIELlemma τίς PronType=Int, and a separate lemma τὶς for the indefinite, as PRON or DET

Mova keeps the accent in the text. An accent-stripping pre-processing, common in Greek NLP for noisy OCR, loses these distinctions and should be flagged as lossy.

13. Sources

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/languages/grc/02-accent/