the assumption of responsibility for the welfare of the world













For translation of an unfamiliar word, place the cursor over the word.


The pronunciation of ΕΛΛΗΝΙΚΗ

As with all dialects, spoken ΕΛΛΗΝΙΚΗ underwent change over time. This would include phonological change, which is the primary evolution for which there is no direct evidence. But comparative evidence and above all reason can approximate and has approximated the pronunciation of various subdialects of ΕΛΛΗΝΙΚΗ at various points in time.

The work of other scholars has established certain facts about the phonology which I take as essentially settled. My contribution to other points is to apply reason where, it seems to me, reason has been lacking. There is no chance of proof, hence no hope of ending controversies, but there is room for further analysis.

It should be noted that the end result of this analysis is an essential pronunciation, a hypothetical reconstruction that includes the most representative traits of the diachronic phonology. It is not necessarily the pronunciation of any one subdialect at any one point. But given that most scholars use a single standard of pronunciation, independent of the point of origin of the text, it would be of value to select as that standard the pronunciation that is most likely to be accurate to the original, across various contexts.

A certain degree of reciprocating evidence is available from the comparative and reconstructive processes in Indo-European linguistics. And scholars have made good use of the interchange among Mediterranean and Near Eastern cultures, and cross-analysis of their writing systems, to establish probabilities for ΕΛΛΗΝΙΚΗ pronunciation. But there has been a serious neglect of what must be the central evidence, which is the ΕΛΛΗΝΙΚΗ orthography itself. And a principle of reason has been overlooked. As with English, the ΕΛΛΗΝΙΚΗ orthography did eventually become fixed, while the pronunciation of the spoken language evolved. But this was a late development, a corruption of the original phonemic, even phonetic, method, which can be seen at work in the early stages of alphabetical writing.

The most reasonable pronunciation of ΕΛΛΗΝΙΚΗ would be that which realized each grapheme as a single phone. The ΕΛΛΗΝΕΣ were unusual in their fidelity to phonetic realizations, and altered their spellings regularly to reflect phonological change. Only two phonetic facts were not consistently recorded: devoicing (rough breathing), and vowel length. But as they were recorded at least inconsistently, these two facts are now generally available.

The stop system is simple to reconstruct and describe; it was regular and complete. There were three points of articulation ― velar, dental, and bilabial ― and four manners of articulation ― nasal, voiced, (unaspirated) voiceless, and aspirated voiceless. The series would be these:

Aspirated Voiceless




As for the other consonants, Λ should always be pronounced as [l], Ξ always as [ks], Ψ always as [ps]. The most characteristic pronunciations of Σ and Ζ were [s] and [zd], but clearly these two graphemes varied across the subdialects, often representing [z] and [dz] respectively. Ρ was a trill, [r], but was devoiced at the beginning of words.

A vowel is a phoneme capable of carrying syllabic length and emphasis. In ΕΛΛΗΝΙΚΗ, a vowel was also the phoneme which was capaple of carrying accent.

There were five vowel phonemes:


low, mid, lax
mid, front, lax
high, front, tense
mid, back, lax, rounded
high, back, tense, rounded

The writing system was only approximate for these sounds. Vowel length, as established through metrical use (in poetry), was inconsistently marked. And the grapheme Ι could and frequently did represent a consonant [j] ― same point and manner of articulation, but without syllabic length. The corresponding consonant for Υ appeared, in classical ΕΛΛΗΝΙΚΗ, only after a doubled vowel; but older texts show the consonant in wider distribution, written F.

The vocalic system was particularly musical, in two ways. The first has to do with rhythm and length. ΕΛΛΗΝΙΚΗ speech can be thought of as a succession of vowels, each carrying one beat (mora). As a rule, each written vowel letter represented one beat. But a vowel could be repeated, that is, doubled in length (forming the so-called long vowels), and yet be written with one letter. In the classical orthography, Η and Ω always carried two beats, Ε and Ο always one. But Α carried one or two beats, and Ι and Υ one, two, or none, and these were not distinguished (with the exception of the Ι subscript, which was always consonantal and should be retained). The diphthongs should be pronounced as written, as a succession of two vowel sounds, each carrying one beat. ΟΥ then was not [u] or even [u:] but Ο followed by Υ. The exception again involves Ι: the sequences ΑΙ and ΟΙ might be diphthongs, but as ultimae they were usually metrically short, which means that the first element carried one beat, and Ι was a consonant.

The second musical quality of ΕΛΛΗΝΙΚΗ vowels was pitch. The so-called accents referred to tone patterns. There were three tones in ΕΛΛΗΝΙΚΗ: normal, high, and low. The acute accent represented a vowel pronounced high, the grave accent a vowel pronounced low, and the circumflex accent a vowel of two beats, the first high, the second normal. Some examples:

ΔΗΜΟΣ, pronounced ΔΕ-Ε-ΜΟΣ; three beats: high-normal-normal.
ΣΩΚΡΑΤΗΣ, pronounced ΣΟ-Ο-ΚΡΑ-ΤΕ-ΕΣ; five beats: normal-normal-high-normal-normal.
ΚΑΙ ΣΥ ΤΕΚΝΟΝ, pronounced ΚΑJ ΣΥ  ΤΕΚ-ΝΟΝ; four beats: low, high, high-normal.
Fifteen beats: high, normal, high-normal, high-normal, low, normal-normal-high-normal, low-low, high-normal.


The pronunciation of LATINA

Less perhaps needs to be said about LATINA. But it is useful to point out here that, rhythmically, LATINA had the same properties. There were five vowels, slightly different from those in ΕΛΛΗΝΙΚΗ:


low, mid, lax
mid, front, tense
high, front, tense
mid, back, tense, rounded
high, back, tense, rounded

LATINA also had the metric system of beats or morae. Unlike in ΕΛΛΗΝΙΚΗ, though, a doubled vowel was nearly always written with a single letter, and so each letter could carry one or two beats. I and V could both function as pre-vocalic consonants (in later times, the consonant/vowel distinction was resolved by J/I and V/U).

Again the principle of reason dictates that the choice of orthography was originally a sensible one. Thus, every letter should be pronounced distinctly, and the combinations AE and OE should be pronounced as A and E, and O and E, not AI and OI.

I am persuaded that LATINA before the classical period had only aspirated voiceless stops (C, T, P), as in English and Deutsch. But by the classical period that had clearly changed, and LATINA shared (as indeed it borrowed) the ΕΛΛΗΝΙΚΗ distinction, unaspirated voiceless stops (C, T, P) and aspirated voiceless stops (CH, TH, PH). The latter, though, served only for borrowings.

I am also persuaded that G before N (as in MAGNVS) represented a velar nasal. For different reasons, the same is true of N before G or C.



The general principle of reproducing language in its original or native form holds. But the writing systems for both ΕΛΛΗΝΙΚΗ and LATINA did evolve, and it is not practical or particularly necessary to match the synchronic form of the graphemes in each era. Today we recognize the equivalence of two main paradigms and countless variations for each unit of the orthography. Though the ancient texts were written in a single case (capitals, hence the Hellenistic alphabet), the transliteration of those graphic forms into equivalents does not subtract much information. The scripts can be substituted by the lowercase, or even the modern mixed system. The addition of spaces and other punctuation, and of diacritical marks for pronunciation, can be helpful without interfering in the reproduction. But for formal circumstances, that reproduction is best which most closely matches the original. At the very least, this would call for unadorned capitals. In LATINA, I and V should be used in preference to J and U, and C should be used in preference to G if done so in the original. It is imperative that ΕΛΛΗΝΙΚΗ writing which included F and Q, used Η for rough breathing, or doubled vowel letters, be reproduced with those facts, for otherwise information is lost from the text.



Home of the Stewardship Project
and O.T. Ford