The Archaic Cretan Greek Alphabet

[From Phoenicia to Greece]  [Table of archaic Cretan alphabet of Dreros & Praisos]  [Phonemic Values of Archaic Letters]

Those who study scripts and forms of writing often distinguish between:

* The Arabic abjad has extended the original Semitic abjad of 22 letters to 28 and considerably re-arranged the order of letters. But recollections of the older system is still preserved in the numeric values attached to letters and, as here, in the word ʔabʤad.

The ancient Semitic abjad consisted of 22 letters, each representing a consonant value. Their order appears to have been fixed at a very early date; and the modern Hebrew script still contains only 22 letters, preserved exactly in the old Semitic order.

However, the ancient Semitic names, from which most of the Greek names ultimately derive, are not preserved. Semitic names are known to us only from very late sources. The earliest are from rabbinical texts and from the Septuagint Greek transcription of names before the alphabetically arranged verses of Jeremiah's Lamentations. The Syriac names are very similar. Many of the Ethiopic names differ considerably. The Arabic names, although derived from older Semitic forms, are considerably shortened and some have clearly been remodelled. In short, the ancient names, are now lost.

By comparing the names of the letters handed down in the various Semitic languages, T. Nöldeke (Beiträge zur semitischen Sprachwissenschaft, 1904, pp. 133 sq.) tried to arrive at the earliest forms. It is these names which I use in these pages, thus:
᾽alf, bēt, gaml, delt, hē, wau, zai, ḥēt, ṭēt, jōd, kaf, lamd, mēm, nūn, semk, ῾ain, pē, ṣādē, qōf, rōš, šīn, tau
Note: 'j' is used with its Italo-Germanic value, i.e. IPA /j/, as in "hallelujah".

From Phoenicia to Greece (from abjad to alphabet):

At sometime in the early years of the 2nd millennium before the the 8th century BCE, the 22 letter Phoenician abjad came to be used to spell Greek and, in the process, became an alphabet. It is most likely to have happened in bilingual trading communities and, indeed, the first to write Greek in Phoenician letters may well have been Phoenician scribes. One such place was certainly Crete.

One extra letter

The most ancient Greek alphabets were content merely to add just one more letter to the Phoenician abjad by, in fact, deriving two letters from the one Phoenician letter wau, namely digamma, which retained its original position as the 6th letter and denoted the consonant /w/, and early u, which took up the 23rd position and denoted the vowels /u/ and /uː/.

'Matres lectionis'

Although the Semitic letters originally all denoted consonant sounds, early on the need was felt, at least where a long vowel had developed following the loss of a final consonant, to preserve the consonant letter as an indication of the vowel. A letter used this way is traditionally termed a mater lectionis "reading mother". The letters most often occurring as matres lectionis were ᾽alf, wau and jōd indicating /aː/, /uː/ and /iː/ respectively. We have seen that wau became two letters, one to denote the consonant /w/ and the other the vowel /u(ː)/. As Greek did not have phonemic /ʔ/ or /j/, ᾽alf and jōd simply acquired the vowel sounds /a(ː)/ and /i(ː)/ respectively. Indeed, it is unlikely that the Greeks even "heard" /ʔ/ and /j/, the consonant values of ᾽alf and jōd, as distinct phonemes.

Glottal and pharyngeal consonants

Greek did not have the range of glottal and pharyngeal sounds that the Semitic languages had; indeed, while the ancient Semitic abjad had letters for four such sounds, some Greeks had only one and others had none! It is unlikely that the Greeks heard the differences as different sounds and in using them as vowels probably thought they were more or less following Phoenician usage. Indeed, there is evidence most may have already functioned as matres lectionis in Semitic usage. The four are:

  1. ᾽alf alpha whose sound was /ʔ/ (voiceless glottal plosive). This is dealt with in the previous paragraph.
  2. epsilon whose sound was /h/ (voiceless glottal fricative). This is used in some early Hebrew texts as a mater lectionis. Its name was obviously heard simply as [eː] and that was the name the early Greeks gave the letter*. They used it to represent /eː/, /e/ and, in some dialects, /ɛː/ also.
    * The name ἔψιλον (epsilon) is a much later invention.
  3. ῾ain O whose sound was /ʕ/ (voiced pharyngeal fricative). There is no reason to suppose the ancient Greeks heard this as a distinct sound any more than most anglophones do today. It was used in the archaic Greek alphabets for both /o/ and /oː/ as well as /ɔː/. According to Hans Jensen (Sign Symbol and Script, London, 1970, p. 457), Hans Bauer, one of the early decipherers of the Ugaritic script, has pointed out that in the Ugaritic cuneiform abjad, ῾ain was apparently used as a mater lectionis for /oː/.
  4. ḥēt eta whose sound was /ħ/ (voiceless pharyngeal fricative). Speakers of western Greek and other dialects, which retained the ancient Greek /h/, used this letter for their /h/, calling it "hēta". But those dialects, such as most of the Ionian dialects and Cretan Doric, that "dropped their aitches" simply called it "ēta"and used it from the start to denote the long low-mid vowel /ɛː/. This was always the usage in the archaic Cretan alphabets.

The 'emphatic' plosives

The old Semitic adjad contained two so-called 'emphatic' plosive consonants:

  1. ṭēt theta whose sound is coventionally transcribed as ṭ and was probably pronounced /t̴/ (a velarized or phayngealized t). This was used in the archaic Greek alphabets and all later alphabets, simplified as Θ, to denote /tʰ/. This is found in the archaic Greek inscriptions at Dreros. It has not been found in any extant Eteocretan texts written in the archaic alphabet.
  2. qōf Ϙ whose sound was /q/. In the earliet Greek inscriptions, Κ was used before α, ε and ι, and Ϙ before ο and υ. But usage varied considerably in local alphabets; eventually Ϙ was dropped everywhere, Κ being used exclusively. It lived on, however, as a numeral for 90 and as Q in the Roman alphabet. It is not found in the archaic Dreros and Praisos inscriptions.

The sibilants

Greek also did not have the range of sibilant consonants that the Semitic languages had, and the use and naming of these consonants shows confusion on the part of the Greeks. The old Phoenician alphabet has four sibilants:

  1. zai zeta whose sound was /z/. It was used from the start to represent a sound which varied in the Greek dialects thus: [dd] ~ [zd] ~ [zz] (see 'Phonemic Values of Archaic Letters' below).
  2. semk semk whose sound was /s/. It is found in some of the archaic alphabets as an alternative way of writing zeta. It was later used in eastern Ionian alphabets to denote /ks/, but this usage is not attested in any of the archaic alphabets nor known in the alphabets of the western Greeks. The letter occurs in the Praisos #1 inscription and is discussed in the next two sections below.
  3. ṣādē san whose sound was probably /s̴/ (velarized or pharyngealized s, i.e. "emphatic" s). This was used in Crete, Thera, Melos, Sikinos, Corinth, Korkyra (Corcyra, Corfu), Sikyon, Argolis and Lokris to denote Greek /s/. The Doric name σάν (san) suggests a conflation of the Semitc ṣādē and šīn.
    Normally, where this sign was adopted, šīn was simply not used. san is regularly used to denote /s/ in the archaic inscriptions at Dreros and Praisos.
  4. šīn Σ or Ϟ whose sound was /ʃ/. This was used in Athens, Euboia, Elis, Lakonia and generally in the later western and eastern Greek alphabets for /s/. Where Σ is used, generally san is not found; but, according to C.D. Buck (The Greek Dialects, Chicago, 1955, p.349) both letters have been found at Argolis and Lokris.
    The Greek name σίγμα or σῖγμα (both accentuations are found in codices) seems to be a conflation of Semitic semk and šīn. The reasons for the confusion of names and distribution of this and the previous letter are unclear. The letter is not used in archaic Cretan alphabets.


Table of archaic Cretan alphabet of Dreros and Praisos

Semitic name᾽alfbētgamldeltwauzaiḥētṭētjōdkaf
Archaic Cretan
alpha beta early gamma, later gamma delta, alternative delta epsilon digamma zeta, (semk) eta theta early iota, later iota kappa
Standard Greek
Modern Roman
Semitic name lamdmēmnūnsemk῾ainṣādēqōfrōššīntau(wau)
Archaic Cretan
lamba mu nu semk o pi san (not
rho (not
tau early u, later u
Standard Greek
Modern Roman

Notes:(The familar English forms of the Byzantine Greek names are given in parentheses)


Phonemic Values of Archaic Letters

Values of the vowels

In the early Greek dialects and, indeed, always in Doric dialects (and Cretan Greek of the historic period was Doric), early u, later u (upsilon) had the high back rounded value /u/, not the high front rounded /y/ of 5th century BCE Athens and later Koine Greek.

As for long and short vowels, only /e/ could have separate symbols for the two quantities, namely eta for "long e" and epsilon for "short e". In fact in Greek spelling it was not even as simple as that. Ancient Greek had two "long e" sounds: low-mid [ɛː] and high-mid [eː]. In Greek eta denoted only the long low-mid sound; epsilon had to do duty for both the short sound and the long high-mid sound (until ει (ei) came to be used to denote /eː/ in the 5th century BCE). We see this in the Cretan Greek inscriptions from Dreros. Whether Eteocretan used similar spelling conventions or whether epsilon was always short, we have no way of knowing; all we can safely assume is that eta is always long.

Omega was a later invention of the Ionian alphabet; no equivalent symbol was used in western Greek alphabets and certainly not in the archaic alphabets: o, whose earliest Greek name appears to have been [oː], had to do duty for both the long and short sounds.

Thus we have no direct evidence whether Eteocretan had long and short values for /i/, /a/, /o/ and /u/. However, the 4th century BCE Praisos #3 inscription indicates some length distinctions and it would be very odd if only one vowel, namely /e/, had long and short quantities while the others did not. We may fairly safely assume, I think, that Eteocretan had the vowels /i/, /e/, /a/, /o/ and /u/ with both long and short quantities.

As for diphthongs, the extant Eteocretan inscriptions are too scant to give us enough information; but /ai/ and /eu/, at least, seem well enough attested. It is probable, I think, that others existed.

Values of consonants

The consonants, for the most part, denoted the same phonemes as those denoted in the International Phonetic Alphabet by the Roman transcription shown in the table above; the only two exceptions are:

  1. zai (zeta): the sound denoted by this symbol seems to have varied in different Greek dialects. Some instances of classical ζ derive from earlier /sd/, e.g. ἵζω (hizo) "I seat" ← *si-sd-ō (cf. Latin: sīdō). The majority of cases, however, derive from a earlier */dj/, */gj/ suggesting that sound denoted by 'z' in transcriptions of Mycenaean Linear B was /dj/ or an affricate such as [ʤ] or [ʣ]. It would seem, however, that in the archaic and classical periods, by a process of assimilation or metathesis, the sound varied in the dialects between [dd], [zd] and [zz] with the latter becoming the norm by the Hellenistic period and giving way eventually to the modern Greek [z].
    In Cretan Greek [dd] was the norm and the spelling δδ is also found. But there appears to have been a tendency in Crete to devoice this combination as ττ is also found for standard Greek ζ; indeed, we also occassionally find actual /tt/ spelled ζ.
    For the above, see: M. Lejeune, Phonétique historique du Mycénien et du Grec ancien, Paris, 1972, pp112 sqq.; W.S. Allen, Vox Graeca, Cambridge UK, 1968, pp. 53 sqq.; C.D. Buck, The Greek Dialects, Chicago, 1955, p. 71 sq., and pp. 313 sqq. However, the letter zeta does not occur in any of the extant Eteocretan texts, so it may not directly concern us here, unless....
  2. semk: semk is found on Praisos #1. As stated above, we can discount the value /ks/ given to this symbol by the Ionians. It would be a gross anachronism to find it used this way in a late 7th century or early 6th century inscription from Crete. There are only two credible possibilities:
    • As in some other local scripts, it is merely used as a variant of zeta and, therefore, presumably denotes either /dd/ or /tt/.
    • It really is semk and is being used to represent a sibilant not known in contemporary Greek. The clear presence of Ϝσ (ws) on Praisos #3 may indicate that Eteocretan possessed a labialized sibilant [sʷ]].
Valid XHTML 1.0 Transitional Valid CSS!
Created August 2003. Last revision:
Copyright © Ray Brown