Little Grammar of Early Modern Greek


These notes on Early Modern Greek grammar are summary and informal: they are intended for proofreaders at the TLG, who are familiar with Classical Greek, and have to proofread Early Modern Greek texts. The notes intend to give them enough information on how the language changed, so that they do not over-correct texts, and make them artificially look more like Classical Greek than they should. For that reason, they aren't exhaustively referenced, and people wanting more should consult the available references on Modern Greek and its history.


There aren't enough references on Early Modern Greek. A grammar is underway in Cambridge, but will not be out for a year or so yet. (And this statement is as true in 2015 as it is in 2009.)

Browning, Tonnet, and Horrocks give a summary introduction to the changes that have taken place, and should be read:

The only comprehensive start-to-finish historical grammar covering Early Modern Greek is Jannaris. It's severely dated, and wrong in places; but until the Cambridge grammar comes out, it's a lot better than nothing: For lexicon, Kriaras covers the period in question, but goes up to σιργ- as of 2015. The abridged Kriaras (which the TLG has digitised) goes only up to παρα-. The abridged Kriaras (eds Kazazis & Karanastasis) is online. Unfortunately, Kriaras definitions are in Modern Greek. As an added inconvenience, Kriaras is monotonic: Kriaras was engaged in the disputes over Greek diglossia, and the presentation of his dictionary reflects a demoticist bias.

Kriaras covers both literary and subliterary texts, but there are some gaps in its coverage; most notably, the Astrological texts of the TLG. Moreover, Kriaras registers phonetic variants of lemmata, but not spelling variants—again, in line with Kriaras' prescriptive bias. So words found in texts may take some hunting in the lexicon.

Trapp (now up to ταρι-) properly covers the middle Byzantine period, but has always spilled over into vernacular texts (the subliterary diplomatic monastic acts in particular); because of the delay in volumes of Kriaras, they have explicitly started covering literary vernacular texts as well.

DuCagne should be too old to be useful; but given the absence of contemporary coverage for the end of the alphabet, it can be used as a back up. The same holds for dictionaries of the contemporary language; I have been importing definitions from the Triantafyllidis Institute's dictionary to supplement the lemmatiser lexicon.

When & What

The changes from Mediaeval to Modern Greek appear to have taken place during the mediaeval Dark Ages—viii to xi AD; and there is little contemporary record of them. The proto-Bulgarian inscriptions, and a few metrical acclamations, date from the period, but still look closer to Middle than Modern Greek. The odd proverb or saying turns up in chronicles or histories towards the end of the period. The first literary text considered to be Early Modern are Michael Glycas' Prison Verses, dated 1158/9; this is a basically learned text, with vernacular proverbs embedded in it. The Ptochoprodromos cycle appears to have begun in the 1170s; but like many early modern texts, probably went through changes before it was first committed to manuscript. This means that all early modern texts' dates of authorship may not be reflected in the texts we actually have.

The end of the period is even more arbitrary. Kriaras uses 1669, the date Candia (Heraklion) fell to the Ottomans; this allows his dictionary to encompass the works of the Cretan Renaissance—though he still includes the works of Cretan refugees, going a decade or so later. But there is no clear linguistic break either side of the fall of Candia.

More crucially, Greek diglossia has always been such that there were few works in a pure vernacular idiom. Being literate in Greek meant being literate in Classical Greek, so all written vernacular works are influenced from the Classical or Byzantine idiom to a greater or lesser degree. This means that the phenomena outlined here will not appear consistently in the texts.

This inconsistency also applies in the contemporary language: the diglossia controversy may have been defused, but it has resulted in a modern standard substantially influenced by Puristic Greek, and frequently in linguistic doublets of vernacular and learned forms, or in the learned form displacing the vernacular. For that reason, Early Modern Greek texts will sound folksy or hypervernacular to Modern Greek speakers, and their judgements on what is linguistically plausible should not always be trusted.


Most Early Modern Greek texts are in verse, and most such verse is in iambic heptameter: "political verse". (The only noteworthy exceptions are the Ptocholeon poem, and Hermoniakos' and Loukanis' retellings of the Iliad, in trochaic tetrameter.) This was also a common verse form for late Byzantine learned verse. Iambic heptameter remained the main verse form up to modern times. The metre matters for accentuation.

The canonical verseform has a strong caesura: — / — / — / — / || — / — / — / — . It allows either hemistich to begin with a choriamb: / — — / — / — / || / — — / — / — .

Until 1500 (but not since then), both learned and vernacular use of the verse also allowed the initial three feet to be replaced with two anapests: — — / — — / — / || — — / — — / — . The unfamiliarity of this variant to Modern Greek speakers has resulted in overemendation in modern editions.

Versifiers are not above misaccenting words metri gratia.

Rhyme is introduced late in the period—the first certain use was in the verse of Stephanos Sachlikes in Crete, 1370s. Rhyme is regular in the Cretan Renaissance, and other Western-ruled domains (Rhodes; there is a Petrarchan sonnet cycle in Cyprus.)

Vernacular metrics does not tolerate hiatus, and elides vowels before other vowels into a single syllable.


Most of the phonetic changes from Ancient to Modern Greek appear to have already been in place by the end of antiquity; the final change, υ, οι /y/ > /i/ , dates from around xi AD. (Michael the Grammarian derides the new pronunciation as provincial in 1030.) (See: F. Lauritzen, Michael the Grammarian's irony about Hypsilon. A step towards reconstructing Byzantine pronunciation. Byzantinoslavica 67, 2009)

Of the stops, aspirates became voiceless fricatives, and voiced stops became voiced fricatives: φ, θ, χ /pʰ, tʰ, kʰ/ > /f, θ, x/, β, δ, γ /b, d, g/ > /v, ð, ɣ/. ζ has long been a single consonant /z/.

Aspiration ceased in post-Classical times; as a result, compounds did not productively aspirate. [This is a matter of ongoing controversy in Greek; "anti-submarine" is αντι-υποβρυχιακός, not ανθ-υποβρυχιακός.]

There are no quantity distinctions for vowels, and there are only three degrees of openness. Diphthongs have been systematically monophthongised; new diphthongs through loans or deletion of consonants are made explicit with diaereses, so diaereses are phonetically significant. So:

The diphthongs ending in upsilon (other than ου) were strengthened from [w] to [f] or [v], depending on whether followed by a voiced or voiceless phoneme:


Modern Greek drastically reduced its permissible final consonants: -ρ, -ψ, -ξ were eliminated, and most final nus became nu movables.

Velar consonants are palatalised before front vowels: /k g x ɣ/ > [c ɟ ç ʝ] (in Cretan and Cypriot [tʃ dʒ ʃ ʒ]). As a consequence, [j] is written as γι "ʝ(i)" initially; ἰατρός > [jatros] γιατρός.

Initial unstressed (/i/, /e/, /o/) were deleted: ἐπιβουλή > πιβουλή, οἰκονόμος > κονόμος, ὀλίγος > λίγος.

Clusters were adjusted to new phonotactic rules: two voiceless fricatives, and two voiceless stops, could not appear in a row. Thus:

Occasionally, these clusters are hypercorrected in the reverse direction, e.g. αὐθίν for αὐτίν. One can even note κθ as a hypercorrection for χθ.

Voiceless stops after nasals became voiced.

The nasal was deleted in some dialects later, and vacillates in the contemporary language. Nasals are deleted before clusters: e.g. μβρ > βρ, νθρ > θρ. Voiced stops in clusters remained stops; because spelling assumed that there was only one pronunciation for β, δ, these were respelled to coincide with the voiced stops above: /i/ and /e/ before vowels is reduced to [j]. It is spelled as an unstressed ι, so that any stress on the yod is transferred to the following vowel: In West Cretan dialect (as found in the Cretan Renaissance), [ja] reduces further to [e]. So μηλέα > Standard Greek μηλιά > West Cretan μηλέ.

The /v/ in αυμ, ευμ is deleted: θαῦμα > θάμα, ψεῦμα > ψέμα.


Modern Greek spelling abandoned graves and breathings on rho by the 1960s, and modern editions do not systematically feature either. Modern Greek has been monotonic (abandoning breathings and circumflexes) since the 1980s, and Early Modern Greek texts are now being printed in the monotonic, with some idiosyncratic use of accent to distinguish minimal pairs.

Modern Greek has no quantity, and this affects accentuation. A "long" ultima is no longer long in anything but spelling; so it can be accented on the antepenult. So if a nominative is proparoxytone, its oblique cases are also proparoxytone: ἄνθρωπος [ˈanθropos] ἄνθρωπου [ˈanθropu]. This does not extend to the genitive plural: ἀνθρώπων [anˈθropon]. (In Cretan dialect, the penult stress extends to the other plural cases: ἀνθρώποι [anˈθropi].)

Likewise, proparoxytone adjectives have proparoxytone feminines: εὔμορφος [ˈevmorfos] εὔμορφη [ˈevmorfi]. This "columnar" stress did not extend to original verb inflections, but it did apply to innovative inflections; ἐρχόμην, but ἔρχουμου, ἔρχομουν. And novel compounds ignore quantity when recessively accented: νεο + νύμφη = νεόνυμφη "new bride".

Enclisis still occurs, but only for proparoxytones. Enclisis is motivated by the three-mora rule, but with no quantity in Modern Greek, properispomena involve just two moras, not three. So while ὁ ἄνθρωπός μου is a real accentuation in Modern Greek, ὁ κῆπός μου is impossible: only ὀ κῆπος μου is possible in the modern language. Likewise, apocope does not result in accentuation of the penult: διατό (> διὰ τό) elides to διατ’, not διάτ’. And if the word contains iota followed by a vowel, it is likely not proparoxytone: κύριε is pronounced in the vernacular [ˈ] and not [ˈkir.i.e], so with enclisis it would be κύριε μου [ˈ] and not κύριέ μου [ˈkir.i.ˈ]. In verse, the metre will make clear which applies.

The enclitics of Modern Greek are changed from the Classical language. Possessive and personal pronouns are still enclitic: 1sg μου με 2sg σου σε 3sg του τον την το 1pl μας 2pl σας 3pl τους. But με meaning "with" is not enclitic but proclitic; the same applies to του τον την το τους when they are articles instead of pronouns. ποτέ in Modern Greek is never enclitic, since it only means "never". τι in Modern Greek is never enclitic, since it only means "what?"

Enclitic and proclitic pronouns may be unaccented. This holds even if they are disyllabic, by the addition of a final -ε: τονε τηνε as variants of τον την "him, her". Otherwise, if a disyllabic grammatical particle is unaccented, it is conventionally accented on the ultima: ὅπου [ˈopu] "where", ὁπού [opu] "relativiser".

Note that learnèd Byzantine Greek is quite arbitrary about accent positioning — to the extent that this should not be treated as an error, but rather is a characteristic of the idiom. This arbitrariness does not extend to the vernacular.

Noun Morphology

The dative had died out in the vernacular by x AD. The dual had died out by the Koine.

Nouns are reduced to first and second declension. In the first declension, the distribution of -η and -α was predictable in Attic; with the dialect mixing of Koine, and loanwords from Latin and subsequent languages, they are no longer predictable, and both occur despite the preceding phoneme: πόλβερη, μπομπάρδα. Because of this, the distinction between contracted first declension and oxytone first declension is artificial.

The regular plural of masculines in the first declension was -αδ- (generalising from -άς -άδος); the feminine ending -ες (itself originally 3rd declension) became general for the masculine only in the 20th century. Late loanwords in -ης formed their plural in -ηδ- instead. Thus, μαθητής has the old plural μαθητάδες, and the contemporary plural μαθητές. The Turkish loanword καδής > kadı has the plural καδήδες. (In older editions, the plural ες is spelled as αις, to maintain some connection to the 1st declension.)

Nouns ending in -ιος were reduced to -ις, and became conflated with first declension -ης: καβαλλάριος καβαλλαρίου > καβαλλάρης καβαλλάρη. Nouns ending in -ιον were reduced to -ι(ν) (occasionally spelled -ην), and formed a distinct declension. Νeuters in -ιό(ν) are rare, and neuters in -ίον are necessarily learned, since the iota should have been reduced to yod.

The Attic second declension was supplanted by normal second declension in the Koine (with Doric forms, if none other were available): e.g. ναός for νεώς.

Third declension nouns were replaced by first declension nouns, with unstressed -ας or -α attached to the noun stem. In the masculine, this gave rise to a novel proparoxytone declension (shown below as -ᾰς). Thus κόραξ, κόρακ-ος > κόρακ-ας, κόρακ-α. Of the third declension noun stems ending in vowels,

Some loanwords from Italian and Turkish end in -ές -έ (masculine) and -έ -ές (feminine). Like -ώ -ώς, these fall into a catch-all declension patterned after the first declension.

The declensions are thus:
1st Decl Fem1st Decl Masc1st Decl Fem Contr1st Decl Masc Contr
PlNomεςεςεςες ~ ᾶδες ~ ῆδεςες ~ ᾶδεςες ~ ᾶδεςᾶδεςᾶδες
Genωνωνωνων ~ άδωνων ~ άδωνων ~ άδωνάδωνάδων
Accες ~ αςες ~ αςες ~ αςες ~ ας ~ ᾶδες ~ άδαςες ~ ας ~ ᾶδες ~ άδας ~ ῆδεςες ~ ας ~ ᾶδες ~ άδαςᾶδες ~ άδαςᾶδες ~ άδας
Vocεςες ~ αςες ~ αςες ~ ᾶδες ~ ῆδεςες ~ ᾶδεςες ~ ᾶδεςᾶδεςᾶδες
2nd Decl Masc2nd Decl Neut
Catch-all MascCatch-all Fem

Adjective Morphology

The major declension of adjectives has three endings: -ος -η -ον. Feminines in -ος are not vernacular. Accent remains columnar: εὔμορφος εὔμορφη εὔμορφο.

Feminine -η becomes -α after iota and epsilon, but not after rho: καρπερός καρπερή καρπερό, πλούσιος πλούσια πλούσιο. This extends to comparatives: χειρότερος χειρότερη χειρότερο.

The -υς -εία (> -ιά) -υ declension survives: μακρύς μακριά μακρύ.

Otherwise, no third declension adjectives survive. No contracted adjectives survive. The -ης -ες declension is not vernacular, and is supplanted by masculine first declension -ης, feminine first declension -α: ζηλιάρης ζηλιάρα. (No neuter is defined: school grammars offer -ικο, but that is properly a distinct inflection.)

Verb Morphology

The dual does not survive. The middle voice does not survive. The 3rd person imperative does not survive. Second aorists do not survive.

Nu movable persists with verb endings, and is even used by analogy where it would not occur in Ancient Greek (or for that matter Standard Modern Greek); e.g. ἐβγάτε(ν) (2nd pl imperative), κεῖνται(ν) (3rd sg present), ἐπερνοῦμα(ν) (1st pl imperf).

The infinitive is dying out during this period, and is restricted to occurring after a few modal verbs. There is some morphological creativity, e.g. εἶσταιν "to be". As elsewhere, the final nu of the infinitive is movable.

The infinitive yields to the subjunctive, which is introduced, in the absence of another conjunction, by να < ἵνα.

The participles survive only in the present active (in an adverbial form, -οντα, later -οντας), and in the perfect passive.

The tense system has collapsed to Present, Imperfect, Aorist. Because of the collapse of vowels into /i/, the distinction between present indicative and present subjunctive is purely orthographical, and has been abandoned in the contemporary orthography, which uses only indicative spellings.

The future, perfect, pluperfect and future perfect are expressed with periphrases. The future was originally expressed as a subjunctive να; from the 1400s it is expressed as θέλω + Infinitive, θέλει(ν) + Subjunctive, or θέλω να + Subjunctive. By 1700, these are reduced to θα + Subjunctive. Because the subjunctive has an aspect distinction (present subjunctive vs. aorist subjunctive), this results in an aspect distinction for futures: future continuous with present subjunctive ("will be doing"), future instantaneous with aorist subjunctive ("will do").

The optative does not survive. Its functions are taken over by να + Imperfect (for optative proper, unreal conditional) or θέλω/θα + Imperfect, ἤθελα + Infinitive, ἤθελε + Subjunctive etc. (for conditional). The development of the θα conditional lags the development of the θα future.

The perfect is expressed from the 1400s on as a combination of ἔχω or εἶμαι, as auxiliary verbs, with the aorist infinitive or passive perfect participle (which does survive). The pluperfect has the auxiliary verb in the imperfect, and lags the development of the perfect. The future perfect (if used) uses the (Modern) future of the auxiliary. So:
Future: OldFuture: NewOptativeConditional
να + Subjunctiveθέλω + Infinitive, θέλω να + Subjunctive, θέλει(ν) + Subjunctive, θα + Subjunctiveνα + Imperfectἤθελα + Infinitive, ἤθελα να + Subjunctive, ἤθελε + Subjunctive
PerfectPluperfectFuture Perfect
Activeἔχω + Active Aorist Infinitive, ἔχω + Passive Perfect Participleεἶχα + Active Aorist Infinitive, εἶχα + Passive Perfect Participleθὰ ἔχω + Active Aorist Infinitive, θὰ ἔχω + Passive Perfect Participle
Passiveἔχω + Passive Aorist Infinitive, εἶμαι + Passive Perfect Participleεἶχα + Passive Aorist Infinitive, ἤμουν + Passive Perfect Participleθὰ ἔχω + Passive Aorist Infinitive, θὰ εἶμαι + Passive Perfect Participle

Athematic conjugations do not survive: εἰμί is reconfigured to a passive εἶμαι. οω-contracts are supplanted by a ων- regular present: δηλόω > δηλώνω. αω-contracts have a hybrid pattern, alternating α and ου for the original αω-contract α and ω, and the εω-contract ει and ου. The εω-contract survives as a distinct pattern, but has been substantially replaced by αω-contracts. Iota subscripts are commonly dispensed with in editions, and are abandoned in the modern orthography.

αω-contracts have analogically uncontracted forms in the singular: -ά-ω, -ά-εις, -ά-ει. The contract passive present also has uncontracted forms, based on εω-imperfect contracts: 3sg imperf act -ειε > 1st pres pass -ειομαι > -ιεμαι. The possible present indicatives are thus:
1sgωῶ, άω
2sgειςᾶς, άειςεῖς
3sgειᾶ, άειεῖ
1plουμεᾶμε, οῦμεοῦμε
3plουσι, ουν(ε)οῦσι, ᾶν(ε)οῦσι, οῦν(ε)
1sgομαιῶμαι, άομαι, ιέμαιοῦμαι
2sgεσαιᾶσαι, άεσαι, ιέσαιεῖσαι
3sgεταιᾶται, άεται, ιέταιεῖται
1plομεθα, ομαστειόμαστειόμαστε

Historically spelled subjunctives substitute ει with ῃ. So the subjunctive can be spelled λύνῃς, λύνης, or λύνεις.

The imperfects are highly variable according to dialect; the following are the standard modern endings, but there is much variation in texts. Note that the personal inflections are merged with the first aorist.
1sgαοῦσα, αγαοῦσα
2sgεςοῦσες, αγεςοῦσες
3sgεοῦσε, αγεοῦσε
1plαμεούσαμε, άγαμεούσαμε
2plατεούσατε, άγατεούσατε
3plαν(ε)ούσαν(ε), άγαν(ε)ούσαν(ε)
1plόμαστε, όμαστανιόμαστε, ιόμαστανιόμαστε, ιόμασταν
2plόσαστε, όσαστανιόσαστε, ιόσαστανιόσαστε, ιόσασταν

The aorist subjunctive remains distinct from the aorist indicative. The aorist passive adds -ηκ-.
3plαν(ε)ουσι, ουν(ε)
3plηκαν(ε)οῦσι, οῦν(ε)

There is no reduplication of the perfect: λυμένος, not λελυμένος. There are no temporal augments: ἀγαπῶ ἀγάπησα. The syllabic augment survives, but may be dropped if unaccented, consistent with the dropping of initial unstressed vowels: (ἐ)λύσανε but ἔλυσαν. There is no internal augment.

The class of productive preverbs is much restricted, with semantic change. μετά- > ματά and ἐξανά- > ξανά- mean "again"; ἐκ- (> ἐξε-) > ξε- means "un-". παρα-, ἀντι-, ἀπο-, κατα- remain productive.

Present stems are frequently remodelled analogically after the aorist. Aorists in -σ- are often mapped to a present in -ν- (especially if they are liquid stems, or end in vowels): so ἔλυσα is mapped back to a present λύνω instead of λύω, and ἔστειλα to στέλνω instead of στέλλω. Presents in -πτω go to -βω. Presents in -σσω go to -ζω. Presents in -ευω often add an epenthetic gamma: -ευγω.


Prepositions only take the accusative. Many ancient prepositions did not survive. σε > εἰς also takes over the role of ἐν; it is merged with the definite article as εἰς τό, εἰς τή > στο, στη "to the, in the". με > μετά means "with"; μετά as an adverb continues to mean "after". Complex spatial relations are expressed by adverbs combined with ἀπό or σε; so "on" is ἐπάνω ἀπό "above from". διά becomes για.

Nick Nicholas, opoudjis [AT] optusnet . com . au
Created: 2009-04-10; Last revision: 2015-02-24