Ancient Greek Phonology

from Wikipedia, the free encyclopedia

The ancient Greek phonology ( Neugriechisch φωνολογία / προφορά της Αρχαίας Ελληνικής γλώσσας fonologia / proforá tis Archéas Ellinikis Glossas ( Dimotiki ) φωνολογία τῆς Ἀρχαίας Ἐλληνικῆς γλώττης fonologia TIS Archaias Ellinikis glottis ( Katharevusa )) is the phonology (or doctrine of pronunciation ) of the ancient Greeks .

First of all, it must be taken into account that the term ancient Greek is problematic as such, as it is a very heterogeneous language form, both in terms of location and over time, which existed in numerous different dialects and was written in texts from Homer to the present day and will. It is clear that there is a not are ancient Greek; rather, there are a multitude of ancient Greek dialects that had different sounds and different pronunciations of the Greek alphabet in different places and at different times . The Attic dialect is commonly referred to as the ancient Greek par excellence, but this harbors the risk of succumbing to the mistaken assumption of a homogeneous ancient Greek language, which never existed, even if the Koine developed since Hellenism . However, since the Attic is the best-known and best-researched ancient Greek dialect due to the many text documents of the Greek classical period, this article is based primarily on the phonology of the Attic.

Three different pronunciation models

There are three different (post-classical) systems according to which ancient Greek texts were or are read aloud:

None of these three pronunciation variants captures the ancient Greek phonology really as ever in a certain place in the Greek-speaking world reality was or how she might also have been only plausible according to current scientific knowledge. Due to the great temporal distance to ancient Greece and the limited number of written documents that have been preserved, there are limits to the full disclosure of ancient Greek phonology. As things stand at the moment, it will never be possible to describe the authentic pronunciation of all ancient Greek dialects at different times with one hundred percent certainty.


In ancient Greek, a distinction was made between long and short vowels. As far as reconstructed, the Attic dialect , which is considered the classical form of Greek, contained five short and seven long vowels. It is difficult to reconstruct their exact pronunciation at a specific time, but the following scheme by W. Sidney Allen (1968) is generally accepted.

Short vowels
Front Back
unrounded rounded unrounded rounded
Closed [ i ] ι / ῐ [ y ] υ / ῠ
center [ e ] ε [ o ] ο
Open [ a ] α / ᾰ
Long vowels
Front Back
unrounded rounded unrounded rounded
Closed [ ] ι / ῑ [ ] υ / ῡ
Half closed [ ] ει ( ε  +  ι ) [ oː ≈ uː ] ου ( ο  +  υ )
Half open [ ɛː ] η [ ɔː ] ω
Open [ ] α / ᾱ

In the case of the a, i and y sounds, the length was not differentiated in the script, the abbreviation or length characters were only introduced in modern times and only sporadically and if so, then only in a linguistic context, never in school lessons , related. The e and o sounds had one short and two long phonemes, which were also spelled differently. The short vowels were represented with epsilon and omicron , it is assumed that they were more like half-closed ([ e ], [ o ]), but it is also possible that they also included the half-open allophones [ ɛ ] and [ ɔ ] . With the long vowels, a distinction was made between the open sounds [ ɛː ] and [ ɔː ] (written with Eta and Omega ) and the closed sounds [ ] and [ ]. The sound [ ɛː ] could possibly have been pronounced more openly, so [ æː ].

The long semi-closed vowels [ ] and [ ] had a complex history. In some cases they have developed from the diphthongs [ ei ] and [ ou ], which is also indicated by the spellings ει and ου. However, in other cases they were created by a replacement stretching of the short sounds [ e ] and [ o ] in order to compensate for a failed consonant. For example, λυθείς lytheis and λύουσι lyousi go back to * λυθεντς * luthents and * λυοντσι * luontsi . In yet another case, [ ] evolved from a syneresis of <εε> and [ ] from one of <εο>, <οε>, or <οο>, with the unconnected forms preserved in other dialects. When the original diphthongs lost their diphthong pronunciation and became [ ] and [ ] , probably in the pre-Classical period, the spellings <ει> and <ου> were a fairly simple method of writing, regardless of their origin. Where the spellings ει and ου match a former diphthong, they are called "real diphthongs", in other cases "false diphthongs".

The sounds [ y ] and [ ] (like German ü ) originally had the sound value [ u ] and [ ]. It is difficult to say with any accuracy when this phonetic shift occurred. It did not occur in all Greek dialects either, but was adopted as the standard by the Koine .

In the post-classical period, the Greek vowels went through numerous changes, which gradually led to the modern Greek system with only five vowel phonemes. During or shortly after the classical period, both [ ] and [ ] shifted and became [ ] and [ ], respectively . [ ] (ει) coincided with the original [ ], while [ ] took the place of the original [ ], which had already shifted to [ ] (see above). The fact that <υ> and <ου> were never confused suggests that the shift of <υ> occurred before that of <ου>, or that the sound shifts occurred in parallel. The rounding from [ y ] to [ i ] occurred in Byzantine times . The distinction between long and short vowels was also given up, so that in the end only the phonemes [a], [ ɛ ], [i], [ ɔ ] and [u] remained.


There were many diphthongs in ancient Greek . All were closing diphthongs, they ended either in [ ] or [ ], as in a semi-vowel final . The first partial could be either long or short, as shown in the table below (the first row of a cell shows the pronunciation in IPA , including its classic spelling):

Front end Back end
Short first sound [ ai̯ ], [ oi̯ ], [ yi̯ ], ([ ei̯ ])
αι, οι, υι, (ει)
[ au̯ ], [ eu̯ ], ([ ou̯ ])
αυ, ευ, (ου)
Long first sound [ aːi̯ ], [ ɛːi̯ ], [ ɔːi̯ ]
ᾱι, ηι, ωι
( ᾳ, ῃ, ῳ )
[ ɛːu̯ ], ([ ɔːu̯ ])
ηυ, (ωυ)

These diphthongs developed differently during and after the classical period (and their partial adoption in the koine). Two of them, ει and ου, were monophthongized early on (see above). The remaining diphthongs that had previously ended on iota also became monophthongs shortly afterwards. This happened early, during or shortly after the classical period. The diphthongs with a long initial sound (<ᾱι>, <ηι> and <ωι>) changed in such a way that the i-sound fell silent and only the initial sound remained: <ᾱ>, <η>, <ω>. The iota was (later) only represented by a mere iota sub- or adscriptum .

The brief diphthongs (<αι>, <οι> and <υι>), however, changed independently: [ ai̯ ] initially shifted to [ ɛː ] and coincided with ε [ Aufgabe ] after the long and short vowels were abandoned . The difference to the Middle Greek pronunciation of the Eta as [ i ] suggests that the shift from [ ɛː ] to [ ] in the case of the Eta before the mentioned shift from <αι> to [ ɛː ] and the following coincidence of Lang- and short vowels (i.e. in the case of the <αι> : [ ɛː ]> [ ɛ ]). <οι> and <υι>, on the other hand, both shifted to [ y ] and later also together with the simple <υ> to [ i ].

The other backward-closing diphthongs (<αυ>, <ευ>, <ηυ>) changed in such a way that the final sound became consonantic during the Hellenistic period, which directly leads to the modern Greek sounds [ av ], [ ev ] and [ iv ] which, however, harden before voiceless consonants and at the end of the word to [ af ], [ ef ] or [ if ] . <ωυ> was rare and only appeared in Ionic , not in classical Attic . Since it was not adopted from the Koine due to its rarity, the digraph no longer appears in today's Greek.


The consonant system of the classical Attic also underwent changes in the post-classical period, but - unlike the vowel system - did not lose any of its complexity. In the system of plosives, for example, there is a tendency towards fricatization . The original plosive sounds were partly formed anew from other sounds, so that there are slightly more consonants in modern Greek, apart from allophones of the actual sounds.


Several sounds pronounced fricative in modern Greek can be assumed with a probability bordering on certainty that they were classical Attic plosives. The ancient grammarians who first tried to classify the sounds, starting with Aristotle , referred to the plosives as ἄφωνα áphōna .

Traditional name Phonetic description Bilabial Alveolar Velar
Ψιλά Psilá , Tenues unvoiced [ p ] π [ t ] τ [ k ] κ
Μέσα Mésa , Mediae voiced [ b ] β [ d ] δ [ ɡ ] γ
Δασέα Daséa , Aspirata aspirated and voiceless [ ] φ [ ] θ [ ] χ

The pronunciation of the original mediae and aspirata changed later : the mediae became voiced ([ v ], [ ð ] or [ ɣ ]), the aspirata became voiceless ([ f ], [ θ ] or [ x ]) fricatives . It is believed that these changes occurred during late antiquity, essentially where koine was spoken, and began with the shift from [ ɡ ] in the third century AD and ended with the aspirata. The Latin transcriptions of <φ>, <θ> and <χ> as "ph", "th" and "ch", which are still used today in foreign words, come from the classic pronunciation of aspirated plosives.

In the case of the labials , the changes must have occurred via the intermediate steps of the bilabial fricatives [ β ] and [ ɸ ], since modern sound values ​​are not bilabial but labiodental .

Other consonants

Besides the plosives, classical ancient Greek knows two nasals ([ m ] and [ n ]), two liquids ([ l ] and [ r ]) and two fricatives ([ h ] and [ s ]), which are dealt with separately in individual sections . Ancient grammarians classified the nasals, liquids and [ s ] as ἡμίφωνα hēmíphōna , by which they probably meant that, in contrast to the áphōna , they could also be pronounced without vowel pronunciation aid . For example, the sound [ b ] is difficult to articulate on its own; it still needs a vowel to generate it, for example [ beː ], which the sound [ ] does not need.

While the terminology of áphōna and hēmíphōna related more to the phonemes than to the letters, the letters Psi (Ψ, ψ), Xi (Ξ, ξ) and Zeta (Ζ, ζ) each stood for consonant connections and were called διπλά diplá ( "Double letters") and - probably because they all contained [ s ] as an element - placed in a group with the hēmíphōna . The pronunciation of the zeta is not entirely clear. For the formation of meters it was treated as a double consonant, so it formed long syllables (see below), but it remains unclear whether it stood for the consonant connection [ zd ] or for [ dz ], or perhaps at different times for one of the two; What is certain is that at the time of its borrowing it was taken over into Latin as [ dzeːta ] (cf. Italian z as [dz] and [ts]). The other two diplá were probably pronounced [ pʰs ] and [ kʰs ] in classical Attic , which can be seen, for example, from the fact that they were written as <φσ> or <χσ> when they were not yet available in Attic as single letters. The aspiration of the first sound was, however, irrelevant for the phonological classification.


There were two nasals in ancient Greek: the bilabial nasal [ m ], written as My (Μ, μ), and the alveolar nasal [ n ], written as Ny (Ν, ν).

Depending on the phonetic environment, the phoneme / n / was realized in speech and writing in four different ways:

  • Before the labials [ b ], [ p ] and [ ] it changes its sound value to [ m ] and is written with μ. For example, the prefix ἐν- en , German 'into' , before βαίνω baínō , -πάθεια -pátheia and φαίνω phaínō is always ἐμ- em- and thus brings the expressions ἐμβαίνω embaíno , German 'I step in, I enter' , ιμ .πάθε empátheia and ἐμφαίνω emphaínō . The same thing happens when an [ s ] follows the plosive in the form of a psi , as in ἔμψυχος émpsychos .
  • Before the nasal [ m ], however, / n / is assimilated or the [ m ] is geminated , and both are pronounced together as an extended bilabial nasal [ ] and written as µµ, for example in ἐν + µένω> ἐμμένω .
  • Before the velars [ ɡ ], [ k ] and [ ], the phoneme / n / was pronounced as [ ŋ ] and written as gamma , for example in ἐγγύς eŋgýs , ἐγκαλέω eŋkaléō , ἐγχέω eŋchéō . The same happens when the velar [ s ] follows in the form of an Xi , as in συγξηραίνω syŋxeraínō , but this occurs less often. The spelling γγ did not stand for the gemination and is therefore not pronounced [ ɡː ].
  • In all other cases the phoneme / n / is pronounced as [ n ], which is also the default.

If possible, the phoneme / n / also enters into geminations itself without being assimilated, for example in the word ἐννέα ennéa , pronounced [ enːe˦a˧ ]. Artificial gemination for metric purposes is also sometimes found, for example in the form ἔννεπε énnepe , pronounced [ e˦nːepe ].


In ancient Greek there were two liquids [ l ] and [ r ], which were written with lambda (Λ, λ) and rho (Ρ, ρ), respectively. If there is an [ n ] in front of the [ l ], a gemination takes place and the combination is pronounced [ ], as in συλλαμβάνω syllambáno , which is based on * συνλαμβάνω * sunlambáno .

The rho probably stood for an alveolar rolled sound, [ r ], as in Italian or today's Greek and regionally in German, rather not for the allophones of English or French. At the beginning of the word the ρ is partly written with Spiritus asper (ῥ), probably to represent an unvoiced or aspirated allophone of [ r ], probably [ ] or [ ], which is where the traditional transcription “rh” comes from. The same spelling is sometimes used when there is a gemination of [ r ], as in συρρέω syrréo , which is therefore sometimes also spelled συῤῥέω syrrhéo , which shifts the transcription to “rrh”. This example also shows that [ n ] in the prefix συν syn is assimilated by a following [ r ], which leads to gemination.


Before the mediae and aspirata became fricatives, Greek likely had only two fricatives: the sibilant [ s ], which was spelled with sigma (Σ, σ, ς), and [ h ]. The former probably had the voiced allophone [ z ] before other voiced consonants, but the spelling did not take this into account, but there is no evidence.

Demonstration of the appearance of using the glyph Ͱ for [h] using the example ἭΡΑ ( Hera )

The sound [ h ] was only at the beginning of the word. In the Attic dialect it was originally written with Heta . Shortly before or during the classical period it fell silent in the Ionian and Aeolian languages , but remained longer in the Attic. In Ionic, the Eta was then used as a vowel letter. When the Ionic alphabet was then adopted by the other regions (for example in Athens in 403 BC), [ h ] still had to be reproduced. In some inscriptions it was written with the left half of the (H) eta instead, see the picture opposite. Later grammarians, during the Hellenistic Koine, changed the symbol to Spiritus asper (Greek δασεῖα daseîa ), which they no longer took as a single letter, but as a diacritical mark that stands over an initial vowel letter. Accordingly, they created another diacritical mark called Spiritus lenis (Greek ψιλή psilḗ ), which should make it clear that the word does not begin with [ h ]. In general, the signs were only introduced in the Byzantine period.

The letter Digamma (Ϝ, ϝ) was used in some dialects for the sound [ w ] in the syllable initial sound . This sound trailed in Attic and Ionian (see. Attic οῖνος oinos , originally ϝοῖνος woînos > Latin vinum > German wine ) before the classical period and was only used as a numeral for "six", where he later also by the stigma was replaced. The [ w ] of other Greek dialects and foreign languages ​​was usually written with <β>, later also with <ου>.

Double consonants

Gemination existed in ancient Greek, i.e. double consonants were lengthened, similar to, for metric purposes, in the modern Cypriot dialect. Double consonants do not appear at the beginning and end of a word. φ, θ and χ are not duplicated in the spelling, the letter combinations πφ, τθ and κχ are used as replacements (cf. the proper names Σαπφώ Sapphṓ , and Βάκχος Bákchos ).

A double sigma of most ancient dialects (and in the koine) - σσ - generally appeared in Attic as a double tau (ττ). Some scholars have suggested that this stands for an affricate ([ ] or [ ts ]), but there is no direct evidence for this.


In ancient Greek, the distinction between long and short syllables is very important. The classical meter is also based on it. A long syllable is a syllable whose vowel is either long, diphthong, or which ends in a consonant. If a consonant is between two syllables within a word, it usually belongs to the second, so the preceding syllable with a short vowel is then a short syllable. If two or more consonants, a double consonant (ζ, ξ or ψ) or an elongated ( geminated ) consonant appear between two syllables within a word, the first one belongs to the one before it and lengthens it. Certain combinations of consonants, voiceless plosives plus liquids or nasals (e.g. τρ or κν), are exceptions, as under certain circumstances both consonants are part of the second syllable - a phenomenon known as “ correptio attica ”. Ancient grammarians referred to a long syllable with a short vowel as " θέσει μακρά thései makrá " - "long according to convention", which was later mistranslated into Latin as " positione longa ". A long vowel syllable was called " φύσει μακρά phýsei makrá " - " natura longa ", "naturally long".


In ancient Greek, a syllable of a word usually had an accent; there were, however, minor exceptions. This emphasis was in Ancient Greek - unlike in modern Greek - primarily as a pitch accent ( pitch accent realized). The stressed syllable was pronounced in a higher pitch and, as a secondary attribute, with a higher volume. According to Dionysius of Halicarnassus , the interval was about a fifth . In the standardized polytonic spelling , which was developed in the Hellenistic period, but only became generally accepted in the Byzantine period, when the tonal accent had already given way to a dynamic one, the acute (Greek ὀξεῖα oxeîa ) was used to indicate a single accented syllable . In long vowels and diphthongs, the accent could fall on either half (or mora ) of the syllable. If the accent fell on the first mora, the syllable first had a high tone and then a normal tone. This was marked in the spelling by the circumflex (Greek περισπωμένη perispōménē ). For example, [ ɛː˥˧ ], ie the drop from a particularly high note to normal pitch with a long [ ɛ ], was reproduced with <ῆ>, [ ɛː˧˥ ], ie the same vowel with a rise from normal pitch to one particularly high tone, with <ή>. The same is true for [ eː˥˧ ] = εῖ and [ eː˧˥ ] = εί, [ yː˥˧ ] = ῦ etc. This also explains why the short vowels epsilon and omicron never appear with a circumflex.

The following rules applied to the distribution of the accents:

For the acute
  • Basically, it could fall on the last three syllables of a word.
  • He could only fall on the third from last syllable if the last syllable contained a short vowel.
  • If (e.g. by changing the case and a long ending of the word) the last syllable of a word in which the acute accent is always on the third last syllable, a long vowel or a diphthong, it shifts to the second last (e.g. Nominative ὁ ἄνθρωπος ho ánthropos , German 'der Mensch' , genitive τοῦ ἀνθρώπου toû anthrópou , German 'des Menschen' ).
  • Is a word with acute on the last syllable not before a final punctuation mark ( point , colon , comma , semicolon ) the acute to be Gravis ( βαρεῖα bareîa ). It may be that this is supposed to indicate a sound decrease; however, there is a lack of evidence for this. Schönberger (2016) cites a number of ancient and Byzantine evidence to prove that the grave accent was never an accent mark in ancient times, but merely indicated the unstressed syllable to be spoken in the normal low tone. He starts from a proclise of oxytonic words within phonetic words of ancient Greek, which originally made oxytonic words toneless in synepia.
For the circumflex
  • It can (see above) only stand on long vowels or diphthongs.
  • It can only appear on the last syllable - or on the penultimate syllable, but only if the last one contains a short vowel.
  • If the last syllable of a word in which the circumflex was on the penultimate syllable is given a long vowel or a diphthong, it becomes an acute (e.g. nominative ὁ Δαρεῖος ho Dareîos , genitive τοῦ Δαρείου toû Dareíou , proper name Darius ); the gravis rule for the acute applies as before.

The diphthongs <αι> and <οι>, if they are at the end of a word and are part of an inflectional ending , are usually treated like short vowels (e.g. οἱ ἄνθρωποι hoi ánthrōpoi ). Schönberger (2016) proposes that this presupposes a monophthongic pronunciation of the diphthongic spellings. Compound words are partially emphasized like their components (e.g. οὔτε oúte , ὥσπερ hṓsper )

Arguments and evidence used in the reconstruction

The above information is based on a large body of evidence that linguists and philologists were constantly debating in the 19th and 20th centuries. The following are some of the arguments used in this argument, along with a brief overview of the sources.

Arguments within the Greek

Initial sound-letter assignment

When a language adopts an alphabet font, there needs to be some simple degree of correspondence between the graphemes (letters) and the phonemes of the adopting language, which does not necessarily mean an exact "one-to-one correspondence". This always leads to the same typing errors as long as the pronunciation remains the same. If a phonetic shift occurs over successive generations, the spelling either changes so that this phonetic shift becomes obvious, or it remains conservative so that traditional spelling prevails. In the first case, which can definitely be called the “spelling reform”, the point in time when the reform was introduced also indicates the point in time when the sound was shifted. In the second case, when historical spelling prevails, spelling mistakes made by inexperienced writers become central points that allow linguists to reconstruct sound shifts and their point in time and thus the development of pronunciation over time.

Spelling mistake

If it is found that scribes particularly often confuse two letters, it can be concluded that the two sounds have coincided. This often happened, for example with <ι> and <ει>, shortly afterwards with <υ> and <οι>, with <ο> and <ω> and with <ε> and <αι> and even later with <η> and <ι> and <ει>, which had already coincided.

If you find out that scribes often omit a letter where it would be used in standard spelling, or that they incorrectly insert one where it doesn't belong (see hypercorrection ), one can conclude that the sound it is Letter represented, got lost in pronunciation. This happened very early with the "Spiritus asper" ([ h ]) at the beginning of the word in most forms of Greek. Another example is the occasional omission of the iota subscriptum in long diphthongs (see above ).

Typos are an important source of evidence, but their occurrence is limited. They only prove that the phonetic development occurred in the language of the document, but not that they prevailed in general. Ancient Greek was not a homogeneous or static language, but was divided into many regional and social variants. Many of the linguistic properties that are characteristic of late and modern Greek probably arose in the sociolects of classical Attic, but the old dialects seem to have persisted for centuries.

Onomatopoeic words

Greek literature sometimes contains representations of animal cries in Greek letters (see onomatopoeia ). The most frequently cited example is the "sheep bleating" < βῆ, βῆ >, which is seen as evidence that the beta was pronounced as a voiced bilabial plosive and that the eta was pronounced as a long, unrounded, semi-open front tongue vowel . Onomatopoetic verbs such as " μυκάομαι mykáomai " (see. Latin " mugire ") for the cow mooing, " βρυχάομαι brycháomai " (see. Latin " rugire ") for the roar of a lion or " κόκκυξ kókkyx " (see. Latin " cuculus " ) for the name of the cuckoo suggest that the archaic pronunciation of the long Ypsilon was [u:] before it became [y:].

Morphophonological Aspects

With some developments within words, sounds are subject to regular changes such as ace or dissimilation , which are sometimes taken into account when writing. This can be used to reconstruct the "original" sounds.

  • <π>, <τ> and <κ> are regularly written at the end of a word as <φ>, <θ> or <χ>, if the following word has a "Spiritus asper" above the first letter. This also applies to compounds. Examples: “ ἐφ 'ἁλός ” instead of “ ἐπὶ ἁλός ” or “ καθ' ἡμᾶς ” instead of “ κατὰ ἡμᾶς ”.
  • The Attic dialect is characterized by syneresis : two consecutive vowels are drawn together to form one syllable. For example, in other dialects, unconnected <εα> regularly becomes <η> in Attic. This corroborates the view that the pronunciation of the Eta was closer to the [ ɛː ] lying exactly between [ e ] and [ a ] and not the [ i ] of modern Greek. Correspondingly, the unconnected Ionic sounds <εε>, <οο> ([ ee ], [ ο.ο ]) are written in Attic combined <ει> and <ου>. This suggests their pronunciation as [ ] or [ οː ] - at least in Attic. Later, however, they became [ i ] and [ u ].

Non-standard spellings

Morphophonological changes like those described above are often treated differently in non-standard spellings. This occasionally leads to questions about the representativeness of the literary dialect and allows reconstructions that would not be possible if only one version were available in the literary texts of the standard language. For example, the non-standard correction of a kappa to a gamma at the end of a word if the following word begins with a voiced consonant, or a kappa to a chi at the end of a word before words beginning with spiritus asper suggests that this kappa should [ ɡ ] or [ ] was assimilated.

Metric evidence

The metra used in classical Greek poetry are based on the principle of long and short syllables and can sometimes serve as evidence of vowel length if this is not already clear from the spelling. From the fourth century AD onwards, poetry was written using stress-based metra, from which one can conclude that there was no longer any distinction between long and short vowels and that the polytonic accent was replaced by a pure stress accent .

Evidence outside of Greek

Sound descriptions

Some ancient grammarians tried to write systematic descriptions of the sounds of the language. Occasional comments on the “correct” pronunciation of individual sounds can be found in other authors. Both types of evidence are often difficult to interpret, as the phonetic terminology of the period was often vague and it is often not clear how the forms described relate to those actually spoken by the broader strata of the population.

Important ancient authors are Dionysius Thrax , Dionysius of Halicarnassus and Ailios Herodianos .

Comparison between different dialects

Sometimes the comparison of the standard Attic with the written forms of the other Greek dialects or the humorous transfer of the “strange” dialectal pronunciation (e.g. Spartan Doric in Attic plays) can provide clues about the sound value of certain spellings.

Towards the end of the fifth century BC Chr. Sometimes transcribed Attic authors the Spartan θ with σ, as in phrases such as " ναὶ τὼ σιώ " (instead of Attic θεώ ), " παρσένε ", " ὀρσά " (instead ὀρθή ), " ἀγασώς " (instead ἀγαθούς ) by Aristophanes ( Lysistrata ). In Thucydides, " σύματος " is found instead of θύματος (the latter spelling was found in descriptions of Doric from the fourth century). One can conclude from this that the <θ> of the Spartan Doric was already a fricative (at least before vowels) and that the Athenian authors parodied this, since the Attic kept the plosive longer .


The spelling of Greek foreign words in other languages ​​and loan words in other languages in Greek can be an important clue for pronunciation. But the evidence is often difficult to interpret or untapped. It should be noted that the sounds of loanwords are often not transferred identically to the other language. Where the target language lacks a phoneme that corresponds exactly to one of the source language, this is usually replaced by a similar sounding phoneme of the target language.

Due to the close proximity of the Roman to the Greek culture, Latin has adopted numerous Greek words. It is therefore of great importance for the reconstruction of ancient Greek phonology. First, Greek loanwords, especially technical names and proper names that contained the letter Φ , were transcribed with “p” or “ph”, with which the scribes tried, albeit incompletely, to write a sound that the Latin did not contain. Later, in the first centuries after Christ, spellings with “f” appear for the first time in such loanwords, which indicates that the phi had already become a fricative . In the second century, "P (h) ilippus" is replaced by "Filippus". Around the same time, the "f" began to be used as a substitute for theta, for want of a better choice, from which one can conclude that the sound of the Greek theta had also become a fricative.

To represent certain other Greek words, the Romans added the letters "y" and "z" to the Latin alphabet, which they adopted directly from the Greek. This is important because it shows that the Romans did not have any characters for the sounds of the letters Υ and Griech in Greek, which means that in this case no Latin sound can be used to reconstruct the Greek sounds.

Comparison with older alphabets

The Greek alphabet evolved from the older Phoenician alphabet . It can be assumed that the Greeks used the different Phoenician letters for these similar Greek sounds. However, similar to the loanwords, this interpretation is affected by numerous variables.

Comparison with recent or derived alphabets

The Greek alphabet is the basis of other alphabets, the Etruscan and later the Armenian , Gothic and Cyrillic alphabets. Arguments similar to those in the Phoenician-Greek case can be used in these cases.

For example, the Cyrillic letter В ( We ) stands for the sound [ v ], which confirms that the beta was already pronounced as a voiced fricative in the ninth century, while the new letter Б ( Be ) was reinvented for the sound [ b ] . In Gothic, on the other hand, the letter derived from beta stands for [b], so beta was still a plosive in the fourth century.

Comparison with modern Greek

Any reconstruction of ancient Greek must take into account how the sounds later evolved into modern Greek, and how these changes occurred. In general, linguists assume that the differences between the reconstructed ancient Greek and modern Greek are relatively unproblematic, since the relevant changes (such as the change from plosives to fricatives , the shift from vowels to [ i ] , the loss of the initial [ h ] and changes in vowel lengths and stress systems) are regularly observed in many languages ​​and are relatively easy to explain.

Comparative reconstruction of Proto-Indo-European

Systematic correspondences between the sounds of Greek and those of the other Indo-European languages serve linguists as strong clues for the reconstruction, since such correspondences are seen as strong indications that these sounds must go back to a common sound of the proto-language .

History of the reconstruction of the ancient pronunciation


By the fifteenth century (during the time of the Byzantine Greek Empire), Greek texts were pronounced the same way as contemporary Greek when read aloud. From around 1486 onwards, various scholars (in particular Antonio de Nebrija , Hieronymus Aleander and Aldus Manutius ) condemned this pronunciation as incompatible with the descriptions that had come down to us from ancient grammarians, and suggested an alternative pronunciation.

Johannes Reuchlin (1455–1522), the leading scholar of Western Greek around 1500, had taken over the teaching of Greek from emigrated Byzantine scholars and continued to use modern pronunciation. The only slightly younger Erasmus of Rotterdam (around 1467–1556) wondered whether the ancient Greek pronunciation could have been different. In 1528 he published his De recta Latini Graecique sermonis pronuntiatione dialogus ("Dialogue on the correct pronunciation of Latin and Greek"), a philosophical treatise in the form of a philosophical dialogue in which he developed a new way of pronouncing ancient Greek and Latin . However, it is said that Erasmus has continued to use the traditional system for teaching. The two models quickly became known as "Reuchlinisches" and "Erasmisches", or, according to the characteristic vowel pronunciations, as " Itazist " (or " Iotazist ") or " Etazist " system, respectively , after the names of their first proponents .

The Erasmic Reconstruction was based on a large body of arguments derived from the philological knowledge available at the time. Mainly he fought for a more regular correspondence between letters and sounds and assumed that different letters must have stood for different sounds and the same letters for the same sounds. This led him to the fact that, for example, the different letters, which are all pronounced [ i ] in the Itazi system , must have had different sound values, and that <ει>, <αι>, <οι>, <ευ>, < αυ> and <ου> must all have been diphthongs with a closing end . He also insisted that the ancient grammarians be taken into account, for example when they prescribed that vowels be of different lengths and shorts, or that the acute and circumflex accents each correspond to a characteristic pitch contour .

He also formulated a number of etymological comparisons with equivalent words from Latin and other European languages. Some of his arguments are, in retrospect, wrong, because the time still lacked the linguistic knowledge. Erasmus therefore did not make a clear distinction between Latin-Greek word relationships that had arisen through borrowing (e.g. Φοῖβος and Latin Phoebus) and those that arose from a common Indo-European root (e.g. φῶρ and Latin furus); in some cases he was also a victim of so-called “ false friends ” and thus only put words together on the basis of coincidental similarities (e.g. Greek θύειν thýein “sacrifice” and French tuer “kill”). In other areas his arguments are quite the same as those used by modern linguists, for example when he argues on the basis of inter-dialectal correspondences within Greek that the Eta is a more open e-sound, closer to [ a ], must have been.

Erasmus went to great lengths to assign plausible phonetic values ​​to the individual phonemes in his reconstructed system. This was not an easy task as there was a lack of rich and accurate terminology in contemporary grammar theories to express such sound values. To circumvent this problem, he used his existing knowledge of the sound repertoires of living languages, for example he compared his reconstructed <η> with a Scottish "a" [ æ ], his reconstructed <ου> with a Dutch "ou" [ ] and his reconstructed <οι> with a French "oi" [ ] ( mind you in a pronunciation that has become historic today).

He stated that the Greek consonant letters <β>, <γ> and <δ> were the voiced plosives [ b ], [ g ], and [ d ], respectively, while the consonant letters <φ>, <θ> and <χ> the fricatives [ f ], [ θ ] or [ x ] are as in modern Greek (although he stated that this f-sound must have been different from the Latin <f>, possibly by which he meant that it was [ φ ] been).

The reception of Erasmus' ideas by his contemporaries varied. The most prominent of the scholars who spoke out against it was Philipp Melanchthon , a student of Johannes Reuchlin. The debate within humanist circles lasted until the seventeenth century, so the situation remained undecided for centuries until it was decided in favor of the Erasmian model.

19th century

A renewed interest in the topic of the reconstructed pronunciation arose in the 19th century. On the one hand, the new science of historical linguistics , based on the method of comparative reconstruction, showed keen interest in Greek. She quickly discovered that, contrary to all doubt, Greek, along with many other languages, belonged to the Indo-European proto- language . This had major consequences for the reconstruction of the phonological system. At the same time, the ongoing work in philology and archeology brought to light an ever-increasing amount of non-literary and non-classical Greek scripts, for example inscriptions and later also papyri, that deviated from the linguistic standard. These discoveries contributed considerably to the expansion of knowledge about the development of the language. On the other hand, academic life in Greece revived after a Greek state saw the light of day again in 1830, and Greek scholars initially only reluctantly accepted the seemingly alien idea that Greek was pronounced so differently from what they knew .

The work of comparative linguists led to a picture of ancient Greek that Erasmus' model initially more or less confirmed, albeit with some changes. It soon became clear that, for example, the pattern of long and short vowels observed in Greek contained similar opposites in other languages ​​that had successors in modern Greek (cf. Ablaut ); that the Greek <υ> must have been [ u ] earlier , because in all other Indo-European languages ​​there is correspondingly [ u ] (cf. Greek μῦς , Latin mūs); that in many cases <η> was before [ a: ] (cf. Greek μήτηρ , Latin māter); that the Greek | <ου> sometimes stood in words for a lengthened <ο> and therefore would have to partly stand for [ o: ] (the same applies analogously to <ε> and the long [ e: ], <ει>), and so on. As for the consonants, the original plosivity of the aspirates <φ>, <θ> and <χ> ([ ], [ ] and [ ]), as well as the mediae <β>, <δ> and < γ> ([ b ], [ d ] and [ g ]), about which it was found out that they are direct further developments of similar sounds in the original Indo-European (reconstructed * [ ], * [ ] and * [ ] , as well as * [ b ], * [ d ] and * [ g ]) were. It was also recognized that the Spiritus asper at the beginning of the word was mostly a remnant of a * [ s ] (cf. Greek ἑπτά and Latin septem), of which it was assumed that its pronunciation had been weakened to [ h ]. Work was also carried out on the reconstruction of the linguistic background of the meter in ancient Greek, especially in Homer, which threw an important light on the phonological structure of syllables and the accent. Scholars also described and explained the regularities of the development of consonants and vowels in processes such as assimilation or reduplication.

Although comparative scholars could safely demonstrate in this way that a certain state, essentially in accordance with the Erasmic model, had applied at a certain time, and that some changes were made later, during the development of modern Greek, the comparative method could do little about it say when this happened. Erasmus had been particularly anxious to find a pronunciation model that would fit as closely as possible to the written letters, and now it was natural to assume that this reconstructed sound inventory was valid at the time when Greek was being written. For a time it was believed to be the pronunciation that had existed throughout the classical period. Nevertheless, it was very possible that the pronunciation of the living language had started quite early in ancient times to change from the reconstructed system to modern Greek.

In these circumstances, the clues from the new, non-standard inscriptions became particularly important. Critics of the Erasmic system paid particular attention to the systematic pattern of spelling mistakes. These errors showed that writers had problems distinguishing the correct spellings for various words, for example <ι>, <η> and <ει>. This proved that these vowels had already started to collapse in the language of that time. Scholars in Greece were quick to emphasize these findings in order to put the Erasmic system down in general, while some Western European scholars tended to downplay them and dismiss them either as isolated exceptions or as influences from non-Attic, non-standard dialects. It seems, however, that some scholars, motivated by the ideological tendency to view post-classical, especially Byzantine and modern Greek, as the vulgar form of the language, wanted ancient Greek to be preserved in a “pure” form. The resulting debate found its expression, for example, in the works of AN Jannaris (1897) and T. Papadimitrakopoulos (1889) on the contra-Erasmic and F. Blass (1870) on the pro-Erasmic side.

It was not until the 20th century that the work of Georgios N. Chatzidakis , who recognized the results of comparative linguistics, was also widely accepted by Greek scholars. The international consensus reached in the early and mid-20th centuries is evidenced by the work of Sturtevant (1940) and Allen (1968).

Recent developments

Since the 1970s and 1980s, some scholars have attempted a systematic reassessment of the evidence from inscriptions and papyri (Teodorsson 1974, 1977, 1978; Gignac 1976; Threatte 1980, summary of Horrocks 1999). According to their results, many of the relevant phonological changes can be well dated, partly still to the classical period, and the time of the koine can be related to many sound changes. Many of the changes in vowelism are now dated between the fifth and first centuries BC, while those of consonants are believed to be complete around the fourth century AD. Even so, there is still considerable debate over precise dates, and it is still not clear to what extent and for how long different pronunciations coexisted within the Greek-speaking community. The consensus-based view today is that a phonological system roughly equivalent to that reconstructed by Erasmus probably applied during the period of classical Attic literature, but biblical or other post-classical Koine Greek was already pronounced in a way that was pronounced in modern Greek already corresponded in essential points.

Recently, however, there has also been an attempt to completely reject the Erasmic reconstruction by the theologian and philologist Chrys C. Caragounis (1995 and 2004). Based on the inscriptions, Caragounis dates all relevant vowel changes during or even before the classical period. He also advocates early conversion of aspirates and mediaes to fricatives and generally doubts the importance of vowel lengths and the differences between accents in spoken language. These views are currently isolated within this scientific field.

In his study, Christos Karvounis comes to the remarkable conclusion that most of the phonological features of modern Greek already existed in ancient times or were in the process of being created - even if not at the same time in one place. It must be assumed that the phonological diversity was already much greater in classical times and that numerous sounds tended much earlier in the direction of today's modern Greek pronunciation than was previously suspected by research.

See also



  • Friedrich Blass : About the pronunciation of Greek . Weidmann, Berlin 1870; 3., reworked. Ed. 1888.
  • Christos Karvounis : Pronunciation and Phonology in Ancient Greek . Knowledge Book Society, Darmstadt 2008, ISBN 978-3-534-20834-0 .
  • Helmut Rix : Historical grammar of Greek. Phonology and form theory. 2nd Edition. Knowledge Buchgesellschaft, Darmstadt 1992, ISBN 3-534-03840-1 .
  • Eduard Schwyzer : Greek grammar . Volume 1: General Part. Phonology, word formation, inflection . Beck, Munich 1939, new edition. 1990, ISBN 3-406-01339-2 .
  • Axel Schönberger: On the treatment of the accentuation of ancient Greek in selected German representations under critical consideration of Greek sources from the first millennium after Christ . Valentia, Frankfurt am Main 2016, ISBN 978-3-936132-39-7 .


  • W. Sidney Allen: Vox Graeca: the pronunciation of Classical Greek . Third edition. University Press, Cambridge 1987, ISBN 0-521-33555-8 .
  • Chrys C. Caragounis: Development of Greek and the New Testament . Baker Academic, 2004, ISBN 3-16-148290-5 .
  • EM Geldart: The Modern Greek Language In Its Relation To Ancient Greek . Reprint. Lightning Source, 2004, ISBN 1-4179-4849-3 .
  • Geoffrey Horrocks : Greek: a history of the language and its speakers . Addison-Wesley, London 1997, ISBN 0-582-30709-0 .
  • Antonios N. Jannaris : An Historical Greek Grammar Chiefly of the Attic Dialect As Written and Spoken From Classical Antiquity Down to the Present Time. MacMillan, London 1887.
  • Philomen Probert: Ancient Greek Accentuation: Synchronic Patterns, Frequency Effects, and Prehistory. Oxford University Press, Oxford 2006.
  • Philomen Probert: Accentuation in Old Attic, Later Attic, and Attic , in: JHW Penney (Ed.), Indo-European Perspectives: Studies in Honor of Anna Morpurgo Davies. Oxford University Press, Oxford 2004, 277-291.
  • Philomen Probert: A New Short Guide to the Accentuation of Ancient Greek. Duckworth (Bristol Classical Press), London 2003.
  • Andrew L. Sihler: New Comparative Grammar of Greek and Latin . Oxford University Press, Oxford 1995, ISBN 0-19-508345-8 .
  • Edgar H. Sturtevant: The Pronunciation of Greek and Latin . New edition 1997. Ares Publishers, Philadelphia 1940, ISBN 0-89005-087-2 .
  • Sven-Tage Teodorsson: The phonemic system of the Attic dialect 400-340 BC. Acta Universitatis Gothoburgensis, Gothenburg 1974.
  • Sven-Tage Teodorsson: The phonology of Ptolemaic Koine (Studia Graeca et Latina Gothoburgensia) . Gothenburg 1977, ISBN 91-7346-035-4 .
  • Sven-Tage Teodorsson The phonology of Attic in the Hellenistic period . Acta Universitatis Gothoburgensis, Göteborg 1978, ISBN 91-7346-059-1 .
  • Leslie Threatte: The grammar of Attic inscriptions. Volume 1: Phonology. de Gruyter, Berlin 1980, ISBN 3-11-007344-7 .
  • WB Stanford: The Sound of Greek . Univ of California Pr, 1967, ISBN 0-520-01204-6 .


  • Michel Lejeune : Phonétique historique du mycénien et du grec ancien . Reprint 2005 Librairie Klincksieck, Paris 1972, ISBN 2-252-03496-3 .
  • A. Meillet: Aperçu d'une histoire de la langue grecque . Eighth edition. Librairie Klincksieck, Paris 1975, ISBN 2-252-03487-4
  • A. Meillet, J. Vendryes: Traité de grammaire comparée des langues classiques . Fourth edition. Librairie Ancienne Honoré Champion, Paris 1968.
  • Jean-Pierre Guglielmi: Grec Ancien - Book + 4 audio CDs (reconstructed phonology). Assimil, 2003, ISBN 2-7005-1087-9 .


  • Angeliki Malikouti-Drachmann: Η φωνολογία της Κλασικής Ελληνικής . In: A.-F. Christidis (Ed.): Ιστορία της Ελληνικής γλώσσας απο τις αρχές εως την ύστερη αρχαιότητα. Ινστιτούτο Νεοελληνικών Σπουδών, 2001, pp. 386-401.
  • Th. Papadimitrakopoulos: Βάσανος τῶν περὶ τῆς ἑλληνικῆς προφορᾶς Ἐρασμικῶν ἀποδείξεων . Athens 1889.
  • Georgios Babiniotis: (Ιστορική Γραμματεία της Αρχαίας Ελληνικής Γλώσσας, 1. Φωνολογία).
  • Hatzidakis: Academic Studies: The pronunciation of Ancient Greek (Ακαδημαϊκαί Αναγνώσματα: Η προφορά της Αρχαίας Ελληνικής). 1902.

Web links

Individual evidence

  1. ^ B. E. Newton: Meter and Stress in Greek. In: Phoenix. 23 (4), 1969, pp. 359-371.